feat: integrate iroh-gossip for mesh networking, using ALPN for protocol routing and node events to manage gossip topics.

Cargo.toml

@@ -3,7 +3,7 @@ resolver = "2"
 members = [
     "lattice-core",
     "lattice-net",
-    "lattice-store",
+    "lattice-cli",
 ]
 
 [workspace.package]
@@ -15,10 +15,13 @@ license = "MIT"
 # Workspace crates
 lattice-core = { path = "lattice-core" }
 lattice-net = { path = "lattice-net" }
-lattice-store = { path = "lattice-store" }
+lattice-cli = { path = "lattice-cli" }
+
+# CLI
+rustyline = "17"
 
 # Networking (Iroh)
-iroh = "0.95"
+iroh = { version = "0.95", features = ["discovery-local-network"] }
 iroh-gossip = "0.95"
 
 # Cryptography
@@ -32,6 +35,8 @@ prost-build = "0.13"
 
 # Async runtime
 tokio = { version = "1", features = ["full"] }
+tokio-util = { version = "0.7", features = ["codec"] }
+futures-util = "0.3"
 
 # Utilities
 thiserror = "2"
@@ -40,6 +45,9 @@ bytes = "1"
 dirs = "5"
 blake3 = "1"
 hex = "0.4"
+redb = "2"
+uuid = { version = "1", features = ["v4"] }
+chrono = "0.4"
 
 # Testing
 tokio-test = "0.4"

docs/architecture.md

@@ -2,20 +2,38 @@
 
 ## Ideas
 
-- SigChains Ed25519-signed, hash-chained append-only logs per node. Trust via local signature verification.
+**Core:**
-- Log-Based State: KV store derived by replaying entries. Watermarks enable safe log pruning + snapshots.
+- SigChains: Ed25519-signed, hash-chained append-only logs per node.
-- Offline-First: Iroh for networking. Vector clocks identify missing entries on reconnect—converges mathematically.
+- Offline-First: Iroh for networking. Vector clocks identify missing entries on reconnect.
-- Full Replication: All nodes keep all logs until watermark consensus, then prune and snapshot.
+- Full Replication: All nodes keep all logs until watermark consensus, then prune.
+
+**State:**
+- Log-Based State: KV store derived from entries. Watermarks enable pruning + snapshots.
+- Merkle-ized State: state.db as Merkle tree. O(1) sync checks, efficient diffing, light clients.
+- DAG Conflict Resolution: Entries track ancestry. Forks merge on next write. Tips only in state.db.
+- KV Snapshots: Point-in-time snapshots for log pruning, fast bootstrap, time travel.
+
+**Operations:**
+- Atomic Batch Writes: Multiple key updates as single entry.
+- Conditional Updates (CAS): Update only if current value matches expected hash.
+
+**CRDTs:**
+- LWW-Register: Last-writer-wins for single values.
+- LWW-Element-Set: Set with add/remove, element present if add > remove timestamp.
 
 ## Concepts
 
-- Transitive Pairing. Nodes can introduce new nodes to the mesh.
+- Transitive Pairing: Nodes can introduce new nodes to the mesh.
+- Multi-Mesh: A node can participate in multiple meshes (clusters). Each mesh is a group of nodes sharing data.
+- Manifest Store: Joining a mesh means joining a special KV store of type "manifest" that defines the mesh membership. The manifest contains node info (`/nodes/{pubkey}/...`).
 
 ## Stack
 
 - rust
 - iroh
 - prost protocol buffers
+- redb (embedded KV store)
+- rustyline (interactive CLI)
 
 ### Bootstrap
 
@@ -41,10 +59,17 @@ Networking modes:
 - Identified by their Ed25519 public key.
 - Private key stored locally in `identity.key` (not replicated).
 - Node data stored in KV:
-  - `/nodes/{pubkey}/info` = static metadata (name, added_by, added_at)
+  - `/nodes/{pubkey}/name` = display name
-  - `/nodes/{pubkey}/status` = `active` | `dormant` | `disabled`
+  - `/nodes/{pubkey}/added_at` = timestamp when added
+  - `/nodes/{pubkey}/status` = `invited` | `active` | `dormant` (removal deletes keys)
   - `/nodes/{pubkey}/role` = `server` | `device` (optional, hints sync priority)
-- Inviting a node = writing entries to `/nodes/{pubkey}/...`.
+- Peer invitation flow:
+  1. Inviter runs `invite <peer_pubkey>` → writes `/nodes/{peer}/info` + `/status`
+  2. Inviter shares their Iroh NodeId out-of-band (QR code, link, text)
+  3. Invited peer runs `join <inviter_nodeid>` → syncs with inviter
+  4. Sync pulls `/nodes/{self}/info` + `/status` → peer is authorized
+  5. `connect` implicitly adds inviter to peer's `/nodes/*` (mutual awareness)
+- Accepting = syncing. The invited peer discovers authorization by receiving the entries.
 - Liveness: Each node tracks `last_seen` locally (from watermark gossip). UI alerts if a peer hasn't been seen for threshold (e.g., 30 days). User decides to mark dormant/disabled.
 - Status effects:
   - `active`: Normal sync participant, blocks watermark until acknowledged.
@@ -58,11 +83,62 @@ Future:
 
 ### Data Model
 
-- Keys are flat strings using path conventions (e.g., `/nodes/{pubkey}`, `/config/sync/interval`).
+- Multiple KV stores supported, identified by `store_id` (UUID).
-- Prefix queries via string matching (sorted map enables efficient range scans).
+- Keys: Arbitrary byte arrays (`Vec<u8>`), sorted lexicographically.
-- State is computed by replaying `Put`/`Delete` operations from all authors.
+- Values: Arbitrary byte arrays (`Vec<u8>`).
+- Each store defines its own key/value format — applications know their schema.
+- Logs are per `(store_id, author_id)` tuple.
+- State is maintained by tracking the "frontier" (tips) of the causal graph for each key.
 - Entry ordering: by HLC timestamp, then by author ID as tiebreaker.
-- Conflicts resolved by last-write-wins (using the ordering above).
+
+**Sync vs Causality:**
+- Vector Clocks track log coverage ("I have entries from Node A up to seq 50") — syncing files.
+- DAG Parents track data causality ("This value replaces that value") — resolving key conflicts.
+
+#### DAG Conflict Resolution
+
+Instead of simple LWW where newest timestamp blindly overwrites, every entry tracks its ancestry:
+
+**Data Model:**
+- Each entry includes `parent_hashes` — references to the entries it supersedes
+- History forms a DAG (directed acyclic graph), not a linear chain
+- state.db stores only "tips" (heads) of the graph per key
+
+**Life Cycle:**
+
+1. **Write (normal):** New entry points to previous entry's hash as parent. History is a straight line.
+
+2. **Write (concurrent/offline):** Two nodes edit same key independently, both pointing to same old parent. History forks into two branches.
+
+3. **Read (forked):** System sees multiple valid values. Uses deterministic rule (highest HLC, then author_id tiebreaker) to return one "winner". No error thrown.
+
+4. **Merge (healing):** Next write to that key cites both existing branches as parents. Fork merges back to single tip.
+
+**Example: Partial Write (Branch Extension)**
+
+```
+Initial: Heads = {A, B} where A(ts:100), B(ts:105). Read winner = B.
+
+Offline node C wakes up, only knows A (hasn't seen B).
+C writes "v3" with parent = [A].
+
+Result: Heads = {C, B}. Conflict shifted, not resolved.
+        C(ts:110) > B(ts:105), so C wins reads.
+
+       ┌──> [A] ──> [C:110]
+[Root]─┤
+       └──> [B:105]
+
+Later: A synced node writes D with parents = [C, B].
+Result: Heads = {D}. Fork merged.
+```
+
+This preserves B's work even though C never saw it. Naive LWW would lose B forever.
+
+#### Store Consistency Modes
+
+- **Eventually consistent**: Default. Writes accepted locally, sync happens async. Fast, offline-capable.
+- **Strictly consistent**: Writes require quorum acknowledgment before commit. Slower, requires connectivity.
 
 ### Timestamps (Hybrid Logical Clocks)
 
@@ -86,27 +162,80 @@ Authors apply their own entries through the standard receive path to ensure cons
 Each node stores logs as one file per author:
 
 ```
-data/
+~/.local/share/lattice/
-├── identity.key               # Local node's Ed25519 private key
+├── identity.key                            # Ed25519 private key
-├── logs/
+├── stores/
-│   └── {author_id_hex}.log    # Append-only SignedEntry stream per author
+│   └── {store_uuid}/
-└── state.db                   # redb: KV snapshot + vector clocks + indexes
+│       ├── logs/
+│       │   └── {author_id_hex}.log         # Append-only SignedEntry stream
+│       └── state.db                        # redb: KV snapshot + frontiers
+└── meta.db                                 # redb: global metadata (known stores, peers)
 ```
 
-- Logs: Append-only binary files per author, containing serialized `SignedEntry` messages.
+- Logs: Append-only binary files per `(store, author)`, containing serialized `SignedEntry` messages.
-- State DB (redb): Combined KV state, vector clocks, and indexes. Updated as entries are applied.
+- State DB (redb): Per-store KV state and frontiers. Updated as entries are applied.
 
-#### state.db Tables (redb)
+#### state.db Tables (per store, redb)
 
 ```
-Table           Key                     Value                      Purpose
+Table              Key                     Value                      Purpose
 ─────────────────────────────────────────────────────────────────────────────
-kv              String (path)           Vec<u8>                    Replicated key-value data
+kv                 Vec<u8> (key)           Vec<HeadInfo>              Current tips for each key
-vector_clocks   [u8; 32] (author_id)    (u64 seq, [u8; 32] hash)   Track sync state + chain verification
+AUTHOR_TABLE      [u8; 32] (author_id)    (u64 seq, [u8; 32] hash)   Per-author frontier tracking
-entry_index     (author_id, seq)        u64 (offset)               Fast entry lookup by position
+meta               Vec<u8>                 Vec<u8>                    Store metadata (incl. merkle_root)
-meta            String                  Vec<u8>                    System metadata (own_seq, watermark, etc.)
 ```
 
+`HeadInfo: { value: Vec<u8>, hlc: u64, author: [u8;32], hash: [u8;32] }`
+
+Note: KV stores multiple heads per key to support DAG conflict resolution. Reads pick winner deterministically.
+
+#### meta.db Tables (global, redb)
+
+```
+Table              Key                     Value                      Purpose
+─────────────────────────────────────────────────────────────────────────────
+stores             [u8; 16] (UUID)         u64 (created_at_ms)        Known stores
+meta               "root_store"            [u8; 16] (UUID)            Root store ID (opened on startup)
+```
+
+- **Root Store**: The primary/manifest store for this node, auto-opened on CLI startup
+- **Stores Table**: Tracks all stores this node participates in
+- Manifest stores define mesh membership via `/nodes/{pubkey}/...` entries
+- Data stores hold application data
+
+#### In-Memory Structures
+
+- log_frontiers: `HashMap<AuthorId, (seq, hash)>` — rebuilt from log files on startup
+
+### Operation Flow (put/delete)
+
+```
+1. User calls put("/key", value)
+         │
+         ▼
+2. SigChain.create_entry()
+   - Build Entry with parent_hashes (current tips for key)
+   - Sign it → SignedEntry
+         │
+         ▼
+3. Append to log + Gossip (critical path)
+   - Write to author's log file
+   - Update log_frontiers (in-memory)
+   - Broadcast to peers
+         │
+         ▼
+4. Apply to state.db (background)
+   - Update kv heads (merge parent tips into new tip)
+   - Update applied_frontiers
+   - Update merkle_root hash
+```
+
+Fast path (1-3): durable + distributed. Background (4): queryable state.
+
+### Read Flow (get)
+
+`get(key)` reads directly from local state.db. Reads are eventually consistent — if state.db lags behind the log, the read may return slightly stale data.
+
 ### Watermarks
 
 - Nodes gossip their watermarks periodically (throttled).
@@ -114,6 +243,11 @@ meta            String                  Vec<u8>                    System metada
 - All nodes keep all logs (own + others) for redundancy until watermark consensus.
 - Once all peers have acknowledged entries, they can be pruned and replaced by the snapshot.
 - If a node is offline too long, it re-bootstraps with a fresh snapshot when it reconnects.
+- Note: Consider preserving logs longer than required for redundancy — enables time travel (view state at any point in history).
+
+**Pruning and DAG Parents:**
+- If a new entry references a parent that was pruned, accept it only if strictly newer than snapshot timestamp.
+- Snapshots act as the base; entries referencing parents older than snapshot are roots relative to that snapshot.
 
 ### Rich CRDTs (Future)
 
@@ -157,3 +291,26 @@ message MergeOp {
 ```
 
 Recommendation: Use Put/Delete for 90% of data. Add CRDT primitives only when needed (concurrent counters, lists) rather than a scripting language.
+
+## Open Questions
+
+### Permissions
+
+Write permissions are enforceable cryptographically:
+- Every entry is signed by author
+- Nodes verify signature before accepting
+- Manifest defines allowed writers: `/nodes/{pubkey}/role` = `writer` | `reader`
+- Entries from non-writers are rejected
+
+Read permissions are not enforceable:
+- Sharing a store = granting read access
+- Encryption adds a layer but doesn't solve revocation (once you have the key, you can read past data)
+- True revocation is impossible — you can't "unread" data
+
+Practical model:
+- Share store = grant read
+- Write access defined in manifest
+- Read-only nodes replicate and verify but can't contribute entries
+
+Future:
+- Capability-based permissions: Explore finer-grained write access (e.g., per-key or per-prefix permissions) via capabilities. Exact mechanism TBD.

docs/roadmap.md

@@ -0,0 +1,217 @@
+# Lattice Roadmap
+
+## Milestone 1: Single-Node Append-Only Log
+
+**Goal:** A single node can create, sign, and persist entries to its own log. No networking yet.
+
+### Deliverables
+
+- [x] HLC timestamps
+- [x] Node identity (Ed25519 keypair, save/load)
+- [x] Entry signing & verification
+- [x] Log file I/O (append, read, hash verification)
+- [x] SigChain (validate entries before appending)
+- [x] Store (redb) — `kv` + `meta` tables, log replay
+- [x] Interactive CLI: `init`, `put`, `get`, `delete`, `status`, `quit`
+
+### Success Criteria
+
+- Can create a new identity
+- Can append entries to local log
+- Can replay log to reconstruct KV state
+- All operations survive restart
+
+### Multi-KV Refactoring (before M2) ✓
+
+- [x] DataDir → `stores/{uuid}/` subdirectories
+- [x] Store → per-store state.db
+- [x] Log paths → `stores/{uuid}/logs/{author}.log`
+- [x] Proto: Entry has store_id (UUID)
+- [x] CLI → `init`, `create-store`, `list-stores`, `use`
+- [x] meta.db stores table (MetaStore)
+- [x] SigChain → validate entry.store_id
+
+---
+
+## Milestone 1.5: DAG Conflict Resolution
+
+**Goal:** Upgrade store from simple LWW to DAG-based conflict resolution per architecture.md.
+
+### Deliverables
+
+- [x] Proto: Add `repeated bytes parent_hashes` to Entry (for DAG causality)
+- [x] Proto: Add `HeadInfo` message for multi-head storage
+- [x] Store: KV table schema → `Vec<u8> → Vec<HeadInfo>`
+- [x] Store: `apply_entry` → track multiple heads, merge parent tips
+- [x] Store: `get` → deterministic winner (highest HLC, author tiebreaker)
+- [x] Store: `get_heads` → inspect all heads for a key
+- [x] EntryBuilder: `.parent_hashes(...)` method for DAG ancestry
+- [x] CLI: Show conflict indicator when multiple heads
+
+### Success Criteria
+
+- [x] Concurrent writes to same key create multiple heads
+- [x] Reads return deterministic winner
+- [x] Next write citing both heads merges fork to single tip
+- [x] All existing tests still pass (71 tests)
+
+---
+
+## Milestone 1.9: Async Refactor
+
+**Goal:** Prepare codebase for concurrent CLI + network operation.
+
+### Deliverables
+
+**Phase 1: Store Actor (sync)** ✓
+- [x] Store actor pattern: dedicated thread owns Store, receives commands via `std::sync::mpsc`
+- [x] StoreHandle wraps channel sender, keeps current API
+- [x] Validate: CLI works as before with actor
+
+**Phase 2: Async Runtime** ✓
+- [x] Add tokio runtime (`#[tokio::main]`)
+- [x] Migrate `std::sync::mpsc` → `tokio::sync::mpsc`
+- [x] Async CLI using `block_in_place` for sync handlers
+
+### Success Criteria
+
+- [x] CLI still works as before
+- [x] Store operations serialized (no data races)
+- [x] Ready for concurrent network tasks
+
+---
+
+## Milestone 2: Two-Node Sync
+
+**Goal:** Two nodes can sync their logs over the network.
+
+### Deliverables
+
+**Phase 1: Sync Logic (no network)** ✓
+- [x] SyncState with AuthorInfo (seq + hash) for hash-based log resumption
+- [x] `Store::sync_state()` → author-to-seq+hash map from AUTHOR_TABLE
+- [x] `SyncState::diff()` → `Vec<MissingRange>` with from_hash for `read_entries_after`
+- [x] Multi-store sync test: compute diff, fetch entries, apply, verify same state
+
+**Phase 2: Iroh Integration**
+
+*Completed:*
+- [x] Node info in root store on init: `/nodes/{pubkey}/info` + `/status`
+- [x] CLI: `invite <pubkey>` to authorize peers
+- [x] CLI: `peers` to list known nodes (with name/added_at info, sorted)
+- [x] CLI: `remove <pubkey>` to remove a peer
+- [x] Iroh endpoint on startup (same Ed25519 key, mDNS + DNS discovery)
+- [x] CLI: `join <nodeid>` - connects to peer, verifies invited
+- [x] Peer verification via `/nodes/{pubkey}/status` check
+
+*Join Protocol (new→existing):* ✓
+- [x] Proto: `JoinRequest` / `JoinResponse` with store UUID
+- [x] Accept handler sends root store UUID in response
+- [x] Join command creates empty store with received UUID (no writes until sync)
+
+*Sync Protocol (bidirectional):* ✓
+- [x] Proto: `PeerMessage` wrapper with `oneof` for message type discrimination
+- [x] `framing.rs` with `MessageSink`/`MessageStream` using `LengthDelimitedCodec`
+- [x] Proto: `SyncRequest`/`SyncResponse` using `SyncState`
+- [x] `Store::read_entries_after(hash)` to fetch log chunks
+- [x] Accept handler: receive SyncState, compute diff, send missing entries
+- [x] Sync command: receive entries, apply to store via `apply_entry`
+- [x] CLI: `sync [nodeid]` command (syncs with all active peers if no nodeid)
+- [x] After sync: node updates own `/nodes/{pubkey}/info` with hostname
+
+*Cleanup*:
+- [x] Move core logic from cmd_join and cmd_sync out of commands.rs (now in `sync.rs`)
+- [x] Add 'invited' state: invite sets 'invited', peer sets 'active' after sync
+
+*Regressions:*
+- [x] Entry ordering: Per-author streaming is correct (hash chain per author, HLC for cross-author).
+- [x] Multi-head sync fixed: SyncState now tracks HashSet of head hashes per author.
+- [x] Sync entry ordering: Entries sent in HLC order (merge-sort across authors) to ensure causal order.
+- [x] `join_mesh` doesn't populate `node.root_store`: Fixed with `complete_join` method.
+
+### Success Criteria
+
+- Node A writes, Node B syncs, both have same state
+- Works offline-first (sync when connected)
+
+**Post-M2 Refactoring:**
+- [x] Unify `node.rs` from `lattice-cli` and `lattice-core`
+- [x] Move network code to `lattice-net`
+
+---
+
+## Milestone 3: Multi-Node Mesh
+
+**Goal:** N nodes form a gossip mesh for real-time sync.
+
+### Deliverables
+
+**Phase 1: LatticeServer Refactor** ✓
+- [x] `LatticeServer` struct in `lattice-net` wrapping `Arc<Node>` + `Endpoint`
+- [x] Move `join_mesh`, `sync_with_peer`, `sync_all` to `LatticeServer` methods
+- [x] Encapsulate accept loop inside `LatticeServer` (via Router + ProtocolHandler)
+- [x] CLI uses `LatticeServer` instead of raw `Node` + `Endpoint`
+- [ ] Integration test: invite → join → sync end-to-end
+- [ ] Periodic background sync with known peers
+- [ ] Track last sync time per peer
+
+**Phase 2: Gossip Protocol** ✓ (iroh-gossip)
+- [x] Router handles both `lattice-sync/1` and `/iroh-gossip/1` ALPNs
+- [x] `NodeEvent::RootStoreActivated` emitted when root store opens
+- [x] Auto-join gossip topic on root store activation
+- [x] Broadcast local entries to gossip topic on commit
+- [x] Receive gossip entries and apply to store
+- [x] Topic ID via `blake3::hash("lattice/{store_id}")`
+- [ ] Gossip bootstrap peers from `/peers/` (needs Prefix Watch)
+
+**Next: Prefix Watch (reactive store updates)**
+- [ ] `store.watch_prefix(prefix) -> Receiver<WatchEvent>`
+- [ ] `WatchEvent::Put { key, value }` / `WatchEvent::Delete { key }`
+- [ ] StoreActor tracks watchers per prefix, emits on matching put/delete
+- [ ] LatticeServer uses `/peers/` watch to update gossip bootstrap peers dynamically
+- [ ] Enables reactive patterns: config changes, presence, app-level subscriptions
+
+---
+
+## Technical Debt
+
+**Logging**
+- [ ] Replace `println!`/`eprintln!` with `tracing` crate (`tracing::info!`, `tracing::error!`)
+- Standard in Rust async ecosystem, used by Iroh internally
+
+**Lifecycle Management (Zombie Tasks)**
+- [ ] Spawned infinite loops (`spawn_node_event_listener`, `spawn_entry_forward_loop`, gossip receive loop) keep running if `LatticeServer` is dropped
+- [ ] Use `tokio_util::sync::CancellationToken` or keep `JoinHandle`s for graceful shutdown
+
+**Error Handling**
+- [ ] Replace `Result<..., String>` with `anyhow::Result` or define `LatticeNetError` enum
+- String errors make it hard to handle specific failure cases
+
+---
+
+## Future
+
+- offline nodes should not delay sync
+- sync command should transitive sync all peers
+- Gossip:
+  - gossip new entries to peers
+  - backfill missing entries from peers (how do peers notice missing entries?)
+  - snapshots for kv store
+  - prune using consensus watermark
+- remove_peer should be a transactional operation on store
+- Watermark tracking & log pruning
+  - Track minimum confirmed seq per author across all peers
+  - Log pruning: remove entries below watermark
+- Multi-KV-Store sync
+- Optimized sync on join. Only transfer current watermark state, then sync missing entries. This would allow pruning. Might need snapshot support in KV store.
+- Mobile (iOS/Android) clients
+- Key rotation
+- Secure storage (Keychain, TPM)
+- Snapshots for fast bootstrap
+- FUSE filesystem mount
+  - Note: FUSE requires u64 inode numbers → maintain `BiMap<u64, Hash>` in redb
+- Merkle-ized State
+  - state.db as Merkle tree with signed root hash
+  - O(1) sync checks (compare root), efficient binary-search diffing
+  - Light clients: fetch value + Merkle proof, verify without full state
+  - Trade-off: write amplification, requires deterministic tree (Patricia Trie / Merkle Search Tree)

docs/storage.md

@@ -0,0 +1,39 @@
+# Storage Format
+
+## Directory Layout
+
+```
+~/.local/share/lattice/
+├── identity.key              # Ed25519 private key (not replicated)
+├── meta.db                   # Global metadata (redb)
+└── stores/{uuid}/
+    ├── logs/{author}.log     # Append-only SignedEntry stream
+    └── state.db              # Per-store KV state (redb)
+```
+
+## meta.db (redb)
+
+| Table   | Key           | Value              | Purpose                      |
+|---------|---------------|--------------------|------------------------------|
+| stores  | UUID (16B)    | created_at (u64)   | Known stores                 |
+| meta    | "root_store"  | UUID (16B)         | Auto-opened on CLI startup   |
+
+## state.db (redb, per store)
+
+| Table   | Key      | Value       | Purpose                |
+|---------|----------|-------------|------------------------|
+| kv      | String   | Vec<u8>     | Key-value data         |
+| meta    | String   | Vec<u8>     | last_seq, last_hash    |
+
+## Log Files
+
+Each `{author}.log` contains length-delimited `LogRecord` messages:
+
+```protobuf
+message LogRecord {
+  bytes hash = 1;         // BLAKE3 hash of entry_bytes
+  bytes entry_bytes = 2;  // Serialized SignedEntry
+}
+```
+
+Hashes are verified on read; corruption causes `LogError::HashMismatch`.

docs/testing.md

@@ -0,0 +1,74 @@
+# Testing Scenarios (Validation Apps)
+
+These apps test HLC ordering, gossip convergence, and conflict resolution.
+
+## Level 1: Pixel Board (Visual Convergence)
+
+50x50 collaborative grid where users paint pixels.
+
+**Data Model:** `/canvas/{x}_{y}` → `{hex_color}`
+
+**Tests:**
+- Visualize sync disagreements immediately
+- High write volume (log performance)
+- Simultaneous writes (HLC tiebreaker)
+
+**Scenario:** Node A paints all red (offline), Node B paints all blue (offline), connect. Board must be identical on both.
+
+---
+
+## Level 2: Shared Grocery List (LWW Trap)
+
+List with add/check/delete operations.
+
+**Data Model:** `/list/{item_uuid}` → `{ name, status: "needed"|"bought" }`
+
+**Tests:** Exposes LWW weakness (resurrection bug)
+
+**Scenario:**
+1. Alice syncs, sees "Milk", goes offline, marks "bought"
+2. Bob syncs, sees "Milk", deletes it
+3. Reconnect
+
+**Result:** Item either resurrects or vanishes based on timestamp. Forces tombstone pattern.
+
+---
+
+## Level 3: Chat Room (Causal Ordering)
+
+Group chat application.
+
+**Data Model:** `/chat/{channel}/{timestamp}_{node_id}` → `{ msg }`
+
+**Tests:**
+- HLC causal ordering
+- Prefix queries (redb range scans)
+- Gap detection via vector clocks
+
+**Scenario:**
+1. Node A sends "Msg 1"
+2. Node B sees it, replies "Msg 2"
+3. Node C comes online, connects only to B
+
+**Success:** Node C receives "Msg 1" before/with "Msg 2" (transitive sync).
+
+---
+
+## Level 4: Chaos Monkey (Automated Simulation)
+
+Tokio-based simulation harness with in-memory networking.
+
+**Setup:**
+- 5 node threads in one process
+- In-memory network (tokio channels)
+- Chaos monkey randomly: cuts connections, writes random keys, sleeps threads
+
+**Assertion:**
+```rust
+let state_0 = nodes[0].dump_state_hash();
+for i in 1..5 {
+    assert_eq!(state_0, nodes[i].dump_state_hash());
+}
+```
+
+Catches HLC clamping edge cases that manual testing misses.

lattice-cli/Cargo.toml

@@ -0,0 +1,25 @@
+[package]
+name = "lattice-cli"
+description = "Interactive CLI for Lattice"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+
+[[bin]]
+name = "lattice"
+path = "src/main.rs"
+
+[dependencies]
+lattice-core = { workspace = true }
+lattice-net = { workspace = true }
+rustyline = { workspace = true }
+hex = { workspace = true }
+thiserror = { workspace = true }
+tokio = { workspace = true }
+shlex = "1"
+serde_json = "1"
+iroh = { workspace = true }
+prost = { workspace = true }
+chrono = { workspace = true }
+tokio-util = { version = "0.7", features = ["codec"] }
+futures-util = "0.3"

lattice-cli/src/commands.rs

@@ -0,0 +1,79 @@
+//! CLI command handlers
+
+use lattice_core::{Node, StoreHandle};
+use lattice_net::LatticeServer;
+
+/// Result of a command that may switch stores or exit
+pub enum CommandResult {
+    /// No store change
+    Ok,
+    /// Switch to this store
+    SwitchTo(StoreHandle),
+    /// Exit the CLI
+    Quit,
+}
+
+/// Helper to call async code from sync command handlers
+pub fn block_async<F: std::future::Future>(f: F) -> F::Output {
+    tokio::task::block_in_place(|| tokio::runtime::Handle::current().block_on(f))
+}
+
+pub type Handler = fn(&Node, Option<&StoreHandle>, Option<&LatticeServer>, &[String]) -> CommandResult;
+
+pub struct Command {
+    pub name: &'static str,
+    pub args: &'static str,
+    pub desc: &'static str,
+    pub group: &'static str,
+    pub min_args: usize,
+    pub max_args: usize,
+    pub handler: Handler,
+}
+
+/// Get all available commands
+pub fn commands() -> Vec<Command> {
+    let mut cmds = Vec::new();
+    
+    // General CLI commands
+    cmds.push(Command { 
+        name: "help", args: "", desc: "Show this help", 
+        group: "general", min_args: 0, max_args: 0, handler: cmd_help as Handler 
+    });
+    cmds.push(Command {
+        name: "quit", args: "", desc: "Exit",
+        group: "general", min_args: 0, max_args: 0, handler: cmd_quit as Handler
+    });
+    
+    // Node commands (operations on the node)
+    cmds.extend(crate::node_commands::node_commands());
+    
+    // Store commands (raw KV operations)
+    cmds.extend(crate::store_commands::store_commands());
+    
+    cmds
+}
+
+fn cmd_help(_node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    let cmds = commands();
+    let mut last_group = "";
+    for cmd in &cmds {
+        if cmd.group != last_group {
+            println!();
+            println!("[{}]", cmd.group);
+            last_group = cmd.group;
+        }
+        let usage = if cmd.args.is_empty() {
+            cmd.name.to_string()
+        } else {
+            format!("{} {}", cmd.name, cmd.args)
+        };
+        println!("  {:18} {}", usage, cmd.desc);
+    }
+    println!();
+    CommandResult::Ok
+}
+
+fn cmd_quit(_node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    println!("Goodbye!");
+    CommandResult::Quit
+}

lattice-cli/src/main.rs

@@ -0,0 +1,121 @@
+//! Lattice Interactive CLI
+
+mod commands;
+mod node_commands;
+mod store_commands;
+
+use lattice_net::LatticeServer;
+use commands::CommandResult;
+use lattice_core::{NodeBuilder, StoreHandle};
+use rustyline::error::ReadlineError;
+use rustyline::DefaultEditor;
+use std::sync::Arc;
+
+#[tokio::main]
+async fn main() {
+    println!("Lattice CLI v{}", env!("CARGO_PKG_VERSION"));
+    println!("Type 'help' for commands, 'quit' to exit.\n");
+
+    let node = match NodeBuilder::new().build() {
+        Ok(n) => Arc::new(n),
+        Err(e) => {
+            eprintln!("Failed to initialize: {}", e);
+            return;
+        }
+    };
+    
+    // Create LatticeServer (creates endpoint and spawns accept loop internally)
+    let server = match LatticeServer::new_from_node(node.clone()).await {
+        Ok(s) => {
+            println!("Iroh:    {} (listening)", s.endpoint().public_key().fmt_short());
+            Some(s)
+        }
+        Err(e) => {
+            eprintln!("Warning: Iroh failed to start: {}", e);
+            None
+        }
+    };
+    
+    let info = node.info();
+    println!("Node ID: {}", info.node_id);
+    println!("Data:    {}", info.data_path);
+
+    if !info.stores.is_empty() {
+        println!("Stores:  {}", info.stores.len());
+    }
+
+    let mut current_store: Option<StoreHandle> = match node.open_root_store().await {
+        Ok(Some(open_info)) => {
+            if open_info.entries_replayed > 0 {
+                println!("Root:    {} (replayed {})", open_info.store_id, open_info.entries_replayed);
+            } else {
+                println!("Root:    {}", open_info.store_id);
+            }
+
+            node.root_store().await.as_ref().cloned()
+        }
+        Ok(None) => {
+            println!("Status:  Not initialized (use 'init')");
+            None
+        }
+        Err(e) => {
+            eprintln!("Warning: {}", e);
+            None
+        }
+    };
+    println!();
+
+    let mut rl = DefaultEditor::new().expect("Failed to create editor");
+    let cmds = commands::commands();
+
+    loop {
+        let prompt = match &current_store {
+            Some(h) => format!("lattice:{}> ", &h.id().to_string()[..8]),
+            None => "lattice:no-store> ".to_string(),
+        };
+
+        match rl.readline(&prompt) {
+            Ok(line) => {
+                let line = line.trim();
+                if line.is_empty() { continue; }
+                let _ = rl.add_history_entry(line);
+
+                let args = match shlex::split(line) {
+                    Some(a) => a,
+                    None => {
+                        println!("Error: mismatched quotes");
+                        continue;
+                    }
+                };
+
+                let cmd_name = args.first().map(|s| s.as_str()).unwrap_or("");
+
+                match cmds.iter().find(|c| c.name == cmd_name || (cmd_name == "exit" && c.name == "quit")) {
+                    Some(cmd) => {
+                        let cmd_args = &args[1..];
+                        if cmd_args.len() < cmd.min_args || cmd_args.len() > cmd.max_args {
+                            println!("Usage: {} {}", cmd.name, cmd.args);
+                        } else {
+                            match (cmd.handler)(&node, current_store.as_ref(), server.as_ref(), cmd_args) {
+                                CommandResult::Ok => {}
+                                CommandResult::SwitchTo(h) => {
+                                    current_store = Some(h);
+                                }
+                                CommandResult::Quit => break,
+                            }
+                        }
+                    }
+                    None => println!("Unknown: '{}'. Type 'help'.", cmd_name),
+                }
+            }
+            Err(ReadlineError::Interrupted | ReadlineError::Eof) => {
+                println!("Goodbye!");
+                break;
+            }
+            Err(e) => {
+                eprintln!("Error: {:?}", e);
+                break;
+            }
+        }
+    }
+}

lattice-cli/src/node_commands.rs

@@ -0,0 +1,313 @@
+//! Node commands - operations on the node (mesh, peers, status)
+
+use crate::commands::{block_async, Command, CommandResult, Handler};
+use lattice_core::{Node, StoreHandle, PeerStatus, Uuid};
+use lattice_net::LatticeServer;
+use chrono::DateTime;
+use std::time::Instant;
+
+pub fn node_commands() -> Vec<Command> {
+    vec![
+        // Store management
+        Command { name: "init", args: "", desc: "Initialize root store", group: "node", min_args: 0, max_args: 0, handler: cmd_init as Handler },
+        Command { name: "create-store", args: "", desc: "Create a new store", group: "node", min_args: 0, max_args: 0, handler: cmd_create_store as Handler },
+        Command { name: "use", args: "<uuid>", desc: "Switch to a store", group: "node", min_args: 1, max_args: 1, handler: cmd_use_store as Handler },
+        Command { name: "list-stores", args: "", desc: "List all stores", group: "node", min_args: 0, max_args: 0, handler: cmd_list_stores as Handler },
+        Command { name: "node-status", args: "", desc: "Show node info", group: "node", min_args: 0, max_args: 0, handler: cmd_node_status as Handler },
+        // Peer management
+        Command { name: "invite", args: "<pubkey>", desc: "Invite a peer", group: "peers", min_args: 1, max_args: 1, handler: cmd_invite as Handler },
+        Command { name: "peers", args: "", desc: "List all peers", group: "peers", min_args: 0, max_args: 0, handler: cmd_peers as Handler },
+        Command { name: "remove", args: "<pubkey>", desc: "Remove a peer", group: "peers", min_args: 1, max_args: 1, handler: cmd_remove as Handler },
+        // Networking
+        Command { name: "join", args: "<node_id>", desc: "Join an existing mesh", group: "network", min_args: 1, max_args: 1, handler: cmd_join as Handler },
+        Command { name: "sync", args: "[node_id]", desc: "Sync with peers", group: "network", min_args: 0, max_args: 1, handler: cmd_sync as Handler },
+    ]
+}
+
+// --- Store management ---
+
+fn cmd_init(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    match block_async(node.init()) {
+        Ok(store_id) => {
+            println!("Initialized with root store: {}", store_id);
+            println!("Node info stored in /nodes/{}/*", hex::encode(node.node_id()));
+            match block_async(node.root_store()).as_ref() {
+                Some(h) => CommandResult::SwitchTo(h.clone()),
+                None => CommandResult::Ok,
+            }
+        }
+        Err(e) => {
+            eprintln!("Error: {}", e);
+            CommandResult::Ok
+        }
+    }
+}
+
+fn cmd_create_store(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    match node.create_store() {
+        Ok(store_id) => {
+            println!("Created store: {}", store_id);
+            match block_async(node.open_store(store_id)) {
+                Ok((handle, _)) => {
+                    println!("Switched to new store");
+                    CommandResult::SwitchTo(handle)
+                }
+                Err(e) => {
+                    eprintln!("Warning: {}", e);
+                    CommandResult::Ok
+                }
+            }
+        }
+        Err(e) => {
+            eprintln!("Error: {}", e);
+            CommandResult::Ok
+        }
+    }
+}
+
+fn cmd_use_store(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let store_id = match Uuid::parse_str(&args[0]) {
+        Ok(id) => id,
+        Err(_) => {
+            eprintln!("Error: invalid UUID '{}'", args[0]);
+            return CommandResult::Ok;
+        }
+    };
+    
+    let start = Instant::now();
+    match block_async(node.open_store(store_id)) {
+        Ok((handle, info)) => {
+            if info.entries_replayed > 0 {
+                println!("Replayed {} entries ({:.2?})", info.entries_replayed, start.elapsed());
+            } else {
+                println!("Switched to store {}", store_id);
+            }
+            CommandResult::SwitchTo(handle)
+        }
+        Err(e) => {
+            eprintln!("Error: {}", e);
+            CommandResult::Ok
+        }
+    }
+}
+
+fn cmd_list_stores(node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    let stores = match node.list_stores() {
+        Ok(s) => s,
+        Err(e) => {
+            eprintln!("Error: {}", e);
+            return CommandResult::Ok;
+        }
+    };
+    let current_id = store.map(|s| s.id());
+    
+    if stores.is_empty() {
+        println!("No stores. Use 'init' or 'create-store'.");
+    } else {
+        for store_id in stores {
+            let marker = if Some(store_id) == current_id { " *" } else { "" };
+            println!("{}{}", store_id, marker);
+        }
+    }
+    CommandResult::Ok
+}
+
+// --- Info ---
+
+fn cmd_node_status(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    println!("Node ID:  {}", hex::encode(node.node_id()));
+    if let Some(name) = node.name() {
+        println!("Name:     {}", name);
+    }
+    println!("Data:     {}", node.data_path().display());
+    match node.root_store_id() {
+        Ok(Some(id)) => println!("Root:     {}", id),
+        Ok(None) => println!("Root:     (not set)"),
+        Err(_) => println!("Root:     (error)"),
+    }
+    
+    // Count peers using node.list_peers()
+    if let Ok(peers) = block_async(node.list_peers()) {
+        let active = peers.iter().filter(|p| p.status == PeerStatus::Active).count();
+        let invited = peers.iter().filter(|p| p.status == PeerStatus::Invited).count();
+        println!("Peers:    {} active, {} invited", active, invited);
+    }
+    
+    CommandResult::Ok
+}
+
+// --- Peer management ---
+
+fn cmd_invite(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let pubkey_hex = &args[0];
+    let pubkey: [u8; 32] = match hex::decode(pubkey_hex) {
+        Ok(bytes) if bytes.len() == 32 => bytes.try_into().unwrap(),
+        _ => {
+            eprintln!("Invalid pubkey: expected 64 hex chars (32 bytes)");
+            return CommandResult::Ok;
+        }
+    };
+    
+    match block_async(node.invite_peer(&pubkey)) {
+        Ok(()) => {
+            println!("Invited peer: {}", pubkey_hex);
+            println!("  Status: {} (will become active after sync)", PeerStatus::Invited.as_str());
+        }
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+fn cmd_peers(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    let peers = match block_async(node.list_peers()) {
+        Ok(p) => p,
+        Err(e) => {
+            eprintln!("Error: {}", e);
+            return CommandResult::Ok;
+        }
+    };
+    
+    if peers.is_empty() {
+        println!("No peers found.");
+        return CommandResult::Ok;
+    }
+    
+    // Group peers by status
+    let mut by_status: std::collections::HashMap<PeerStatus, Vec<&lattice_core::PeerInfo>> = 
+        std::collections::HashMap::new();
+    for peer in &peers {
+        by_status.entry(peer.status).or_default().push(peer);
+    }
+    
+    // Print grouped by status in order: active, invited, dormant
+    let status_order = [PeerStatus::Active, PeerStatus::Invited, PeerStatus::Dormant];
+    for status in &status_order {
+        if let Some(peer_list) = by_status.get(status) {
+            println!("\n[{}] ({}):", status.as_str(), peer_list.len());
+            let mut sorted: Vec<_> = peer_list.iter().collect();
+            sorted.sort_by(|a, b| a.pubkey.cmp(&b.pubkey));
+            for peer in sorted {
+                let added_str = peer.added_at
+                    .and_then(|ts| DateTime::from_timestamp(ts as i64, 0))
+                    .map(|dt| dt.format("%Y-%m-%d").to_string())
+                    .unwrap_or_default();
+                let info_str = match (peer.name.as_ref(), added_str.is_empty()) {
+                    (Some(name), false) => format!(" {} ({})", name, added_str),
+                    (Some(name), true) => format!(" {}", name),
+                    (None, false) => format!(" ({})", added_str),
+                    (None, true) => String::new(),
+                };
+                println!("  {}{}", peer.pubkey, info_str);
+            }
+        }
+    }
+    CommandResult::Ok
+}
+
+fn cmd_remove(node: &Node, _store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let pubkey_hex = &args[0];
+    let pubkey: [u8; 32] = match hex::decode(pubkey_hex) {
+        Ok(bytes) if bytes.len() == 32 => bytes.try_into().unwrap(),
+        _ => {
+            eprintln!("Invalid pubkey: expected 64 hex characters");
+            return CommandResult::Ok;
+        }
+    };
+    
+    match block_async(node.remove_peer(&pubkey)) {
+        Ok(()) => println!("Removed peer: {}...", &pubkey_hex[..10]),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+// --- Networking ---
+
+fn cmd_join(_node: &Node, store: Option<&StoreHandle>, server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let server = match server {
+        Some(s) => s,
+        None => {
+            eprintln!("Iroh endpoint not started.");
+            return CommandResult::Ok;
+        }
+    };
+    
+    if store.is_some() {
+        eprintln!("Already initialized. Use 'sync' to sync with peers.");
+        return CommandResult::Ok;
+    }
+    
+    let peer_id = match lattice_net::parse_node_id(&args[0]) {
+        Ok(id) => id,
+        Err(e) => {
+            eprintln!("Invalid node ID: {}", e);
+            return CommandResult::Ok;
+        }
+    };
+    
+    println!("Joining mesh via {}...", peer_id.fmt_short());
+    
+    match block_async(server.join_mesh(peer_id)) {
+        Ok(handle) => {
+            println!("Joined mesh! Use 'sync' command to sync entries.");
+            CommandResult::SwitchTo(handle)
+        }
+        Err(e) => {
+            eprintln!("Join failed: {}", e);
+            CommandResult::Ok
+        }
+    }
+}
+
+fn cmd_sync(_node: &Node, store: Option<&StoreHandle>, server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let server = match server {
+        Some(s) => s,
+        None => {
+            eprintln!("Iroh endpoint not started.");
+            return CommandResult::Ok;
+        }
+    };
+    
+    let store = match store {
+        Some(s) => s,
+        None => {
+            eprintln!("No store open. Use 'init' or 'join' first.");
+            return CommandResult::Ok;
+        }
+    };
+    
+    if args.is_empty() {
+        // Sync with all active peers
+        match block_async(server.sync_all(store)) {
+            Ok(results) => {
+                if results.is_empty() {
+                    println!("No peers to sync with.");
+                } else {
+                    let total: u64 = results.iter().map(|r| r.entries_applied).sum();
+                    println!("\nSync complete! Applied {} entries from {} peer(s).", total, results.len());
+                }
+            }
+            Err(e) => eprintln!("Sync failed: {}", e),
+        }
+    } else {
+        // Sync with specific peer
+        let peer_id = match lattice_net::parse_node_id(&args[0]) {
+            Ok(id) => id,
+            Err(e) => {
+                eprintln!("Invalid node ID: {}", e);
+                return CommandResult::Ok;
+            }
+        };
+        
+        println!("Syncing with {}...", peer_id.fmt_short());
+        match block_async(server.sync_with_peer(store, peer_id)) {
+            Ok(result) => {
+                println!("Sync complete! Applied {} entries (peer sent {})", 
+                    result.entries_applied, result.entries_sent_by_peer);
+            }
+            Err(e) => eprintln!("Sync failed: {}", e),
+        }
+    }
+    
+    CommandResult::Ok
+}

lattice-cli/src/store_commands.rs

@@ -0,0 +1,220 @@
+//! Store commands - direct KV operations
+
+use crate::commands::{block_async, Command, CommandResult, Handler};
+use lattice_core::{Node, StoreHandle};
+use lattice_net::LatticeServer;
+use std::time::Instant;
+
+pub fn store_commands() -> Vec<Command> {
+    vec![
+        Command { name: "store-status", args: "", desc: "Show store info", group: "store", min_args: 0, max_args: 0, handler: cmd_store_status as Handler },
+        Command { name: "put", args: "<key> <value>", desc: "Store a key-value pair", group: "store", min_args: 2, max_args: 2, handler: cmd_put as Handler },
+        Command { name: "get", args: "<key> [-v]", desc: "Get value for key", group: "store", min_args: 1, max_args: 2, handler: cmd_get as Handler },
+        Command { name: "delete", args: "<key>", desc: "Delete a key", group: "store", min_args: 1, max_args: 1, handler: cmd_delete as Handler },
+        Command { name: "list", args: "[prefix] [-v]", desc: "List keys (optionally filtered by prefix)", group: "store", min_args: 0, max_args: 2, handler: cmd_list as Handler },
+        Command { name: "author-state", args: "[pubkey]", desc: "Show author sync state", group: "store", min_args: 0, max_args: 1, handler: cmd_author_state as Handler },
+    ]
+}
+
+fn cmd_store_status(_node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, _args: &[String]) -> CommandResult {
+    let Some(h) = store else {
+        println!("No store selected. Use 'init' or 'use <uuid>'");
+        return CommandResult::Ok;
+    };
+    
+    println!("Store ID: {}", h.id());
+    println!("Log Seq:  {}", block_async(h.log_seq()));
+    println!("Applied:  {}", block_async(h.applied_seq()).unwrap_or(0));
+    
+    let all = block_async(h.list(false)).unwrap_or_default();
+    println!("Keys:     {}", all.len());
+    
+    // Show log directory size
+    let (file_count, total_size) = block_async(h.log_stats());
+    if file_count > 0 {
+        println!("Logs:     {} files, {} bytes", file_count, total_size);
+    }
+    
+    CommandResult::Ok
+}
+
+fn cmd_put(_node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let Some(h) = store else {
+        println!("No store selected. Use 'init' or 'use <uuid>'");
+        return CommandResult::Ok;
+    };
+    let start = Instant::now();
+    match block_async(h.put(args[0].as_bytes(), args[1].as_bytes())) {
+        Ok(seq) => println!("OK (seq: {}, {:.2?})", seq, start.elapsed()),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+fn cmd_get(_node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let Some(h) = store else {
+        println!("No store selected. Use 'init' or 'use <uuid>'");
+        return CommandResult::Ok;
+    };
+    let verbose = args.get(1).map(|a| a == "-v").unwrap_or(false);
+    let start = Instant::now();
+    let key = args[0].as_bytes();
+    
+    if verbose {
+        // Show all heads
+        match block_async(h.get_heads(key)) {
+            Ok(heads) if heads.is_empty() => println!("(nil)"),
+            Ok(heads) => {
+                for (i, head) in heads.iter().enumerate() {
+                    let winner = if i == 0 { "→" } else { " " };
+                    let tombstone = if head.tombstone { "⊗" } else { "" };
+                    let author_short = hex::encode(&head.author).chars().take(8).collect::<String>();
+                    if head.tombstone {
+                        println!("{} {} (deleted) (hlc:{}, author:{})", 
+                            winner, tombstone, head.hlc, author_short);
+                    } else {
+                        println!("{} {} (hlc:{}, author:{})", 
+                            winner, format_value(&head.value), head.hlc, author_short);
+                    }
+                }
+                if heads.len() > 1 {
+                    println!("⚠ {} heads (conflict)", heads.len());
+                }
+                println!("({:.2?})", start.elapsed());
+            }
+            Err(e) => eprintln!("Error: {}", e),
+        }
+    } else {
+        match block_async(h.get(key)) {
+            Ok(Some(v)) => {
+                let heads = block_async(h.get_heads(key)).unwrap_or_default();
+                if heads.len() > 1 {
+                    println!("{} (⚠ {} heads)", format_value(&v), heads.len());
+                } else {
+                    println!("{}", format_value(&v));
+                }
+                println!("({:.2?})", start.elapsed());
+            }
+            Ok(None) => println!("(nil)"),
+            Err(e) => eprintln!("Error: {}", e),
+        }
+    }
+    CommandResult::Ok
+}
+
+fn cmd_delete(_node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let Some(h) = store else {
+        println!("No store selected. Use 'init' or 'use <uuid>'");
+        return CommandResult::Ok;
+    };
+    let start = Instant::now();
+    match block_async(h.delete(args[0].as_bytes())) {
+        Ok(seq) => println!("OK (seq: {}, {:.2?})", seq, start.elapsed()),
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+fn cmd_list(_node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let Some(h) = store else {
+        println!("No store selected. Use 'init' or 'use <uuid>'");
+        return CommandResult::Ok;
+    };
+    
+    // Parse args: [prefix] [-v]
+    let verbose = args.iter().any(|a| a == "-v");
+    let prefix = args.iter().find(|a| *a != "-v").cloned();
+    
+    let start = Instant::now();
+    let result = if let Some(p) = &prefix {
+        block_async(h.list_by_prefix(p.as_bytes(), verbose))
+    } else {
+        block_async(h.list(verbose))
+    };
+    
+    match result {
+        Ok(entries) => {
+            if entries.is_empty() {
+                println!("(empty)");
+            } else {
+                for (k, v) in &entries {
+                    let key_str = format_value(k);
+                    if verbose {
+                        // Show all heads for this key
+                        let heads = block_async(h.get_heads(k)).unwrap_or_default();
+                        println!("{}:", key_str);
+                        for (i, head) in heads.iter().enumerate() {
+                            let winner = if i == 0 { "→" } else { " " };
+                            let author_short = hex::encode(&head.author).chars().take(8).collect::<String>();
+                            if head.tombstone {
+                                println!("  {} ⊗ (deleted) (hlc:{}, author:{})", 
+                                    winner, head.hlc, author_short);
+                            } else {
+                                println!("  {} {} (hlc:{}, author:{})", 
+                                    winner, format_value(&head.value), head.hlc, author_short);
+                            }
+                        }
+                    } else {
+                        // Check for multiple heads
+                        let heads = block_async(h.get_heads(k)).unwrap_or_default();
+                        if heads.len() > 1 {
+                            println!("{} = {} (⚠ {} heads)", key_str, format_value(v), heads.len());
+                        } else {
+                            println!("{} = {}", key_str, format_value(v));
+                        }
+                    }
+                }
+                let prefix_str = prefix.as_ref().map(|p| format!(" (prefix: {})", p)).unwrap_or_default();
+                println!("({} keys{}, {:.2?})", entries.len(), prefix_str, start.elapsed());
+            }
+        }
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+fn cmd_author_state(node: &Node, store: Option<&StoreHandle>, _server: Option<&LatticeServer>, args: &[String]) -> CommandResult {
+    let store = match store {
+        Some(s) => s,
+        None => {
+            eprintln!("Error: no store selected");
+            return CommandResult::Ok;
+        }
+    };
+
+    // Get author: from arg or default to self
+    let author_bytes: [u8; 32] = if args.is_empty() {
+        node.node_id()
+    } else {
+        let hex_str = args[0].trim_start_matches("0x");
+        match hex::decode(hex_str) {
+            Ok(bytes) if bytes.len() == 32 => bytes.try_into().unwrap(),
+            Ok(bytes) => {
+                eprintln!("Error: author must be 32 bytes, got {}", bytes.len());
+                return CommandResult::Ok;
+            }
+            Err(e) => {
+                eprintln!("Error: invalid hex: {}", e);
+                return CommandResult::Ok;
+            }
+        }
+    };
+
+    match block_async(store.author_state(&author_bytes)) {
+        Ok(Some(state)) => {
+            println!("Author: {}", hex::encode(&author_bytes));
+            println!("  seq: {}", state.seq);
+            println!("  hash: {}", hex::encode(&state.hash));
+            println!("  log_offset: {}", state.log_offset);
+        }
+        Ok(None) => {
+            println!("No state for author: {}", hex::encode(&author_bytes));
+        }
+        Err(e) => eprintln!("Error: {}", e),
+    }
+    CommandResult::Ok
+}
+
+fn format_value(v: &[u8]) -> String {
+    std::str::from_utf8(v).map(String::from).unwrap_or_else(|_| format!("0x{}", hex::encode(v)))
+}

lattice-core/Cargo.toml

@@ -14,6 +14,11 @@ bytes = { workspace = true }
 dirs = { workspace = true }
 blake3 = { workspace = true }
 hex = { workspace = true }
+redb = { workspace = true }
+uuid = { workspace = true }
+tokio = { workspace = true }
+hostname = "0.4"
+serde_json = "1"
 
 [build-dependencies]
 prost-build = { workspace = true }

lattice-core/src/causal_iter.rs

@@ -0,0 +1,189 @@
+//! Causal Entry Iterator - yields entries in HLC (causal) order
+//!
+//! Implements merge-sort streaming across multiple author queues using a min-heap,
+//! ensuring entries are returned in correct causal order for sync.
+//! Complexity: O(N log K) where N = total entries, K = number of authors.
+
+use crate::proto::{Entry, SignedEntry};
+use prost::Message;
+use std::cmp::Ordering;
+use std::collections::{BinaryHeap, VecDeque};
+
+/// A heap entry that wraps an author queue index and the HLC of its front entry.
+/// Uses Reverse for min-heap behavior (lowest HLC first).
+struct HeapEntry {
+    hlc: (u64, u32),
+    queue_idx: usize,
+}
+
+impl PartialEq for HeapEntry {
+    fn eq(&self, other: &Self) -> bool {
+        self.hlc == other.hlc
+    }
+}
+
+impl Eq for HeapEntry {}
+
+impl PartialOrd for HeapEntry {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for HeapEntry {
+    fn cmp(&self, other: &Self) -> Ordering {
+        // Reverse order for min-heap (BinaryHeap is max-heap by default)
+        other.hlc.cmp(&self.hlc)
+    }
+}
+
+/// Iterator that yields SignedEntry in HLC (causal) order.
+/// 
+/// Takes multiple VecDeques (one per author) and yields entries
+/// from lowest to highest HLC, ensuring causal ordering for sync.
+/// Uses a min-heap for O(log K) per-entry overhead instead of O(K) linear scan.
+pub struct CausalEntryIter {
+    queues: Vec<VecDeque<SignedEntry>>,
+    heap: BinaryHeap<HeapEntry>,
+}
+
+impl CausalEntryIter {
+    /// Create a new iterator from a list of entry queues (one per author)
+    pub fn new(queues: Vec<VecDeque<SignedEntry>>) -> Self {
+        let mut heap = BinaryHeap::with_capacity(queues.len());
+        
+        // Initialize heap with the front entry from each non-empty queue
+        for (idx, queue) in queues.iter().enumerate() {
+            if let Some(entry) = queue.front() {
+                heap.push(HeapEntry {
+                    hlc: Self::get_hlc(entry),
+                    queue_idx: idx,
+                });
+            }
+        }
+        
+        Self { queues, heap }
+    }
+    
+    /// Extract HLC (wall_time, counter) from a SignedEntry
+    fn get_hlc(entry: &SignedEntry) -> (u64, u32) {
+        Entry::decode(&entry.entry_bytes[..])
+            .ok()
+            .and_then(|e| e.timestamp)
+            .map(|t| (t.wall_time, t.counter))
+            .unwrap_or((0, 0))
+    }
+}
+
+impl Iterator for CausalEntryIter {
+    type Item = SignedEntry;
+    
+    fn next(&mut self) -> Option<Self::Item> {
+        // Pop the queue with lowest HLC
+        let HeapEntry { queue_idx, .. } = self.heap.pop()?;
+        
+        // Remove entry from that queue
+        let entry = self.queues[queue_idx].pop_front()?;
+        
+        // If queue still has entries, push its new front back to heap
+        if let Some(next_entry) = self.queues[queue_idx].front() {
+            self.heap.push(HeapEntry {
+                hlc: Self::get_hlc(next_entry),
+                queue_idx,
+            });
+        }
+        
+        Some(entry)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::hlc::HLC;
+    use crate::clock::MockClock;
+    use crate::node_identity::NodeIdentity;
+    use crate::signed_entry::EntryBuilder;
+    
+    fn make_entry(node: &NodeIdentity, seq: u64, clock_ms: u64) -> SignedEntry {
+        let clock = MockClock::new(clock_ms);
+        EntryBuilder::new(seq, HLC::now_with_clock(&clock))
+            .store_id(vec![0u8; 16])
+            .prev_hash(vec![0u8; 32])
+            .put(b"/test".to_vec(), format!("seq{}", seq).into_bytes())
+            .sign(node)
+    }
+    
+    #[test]
+    fn test_empty_iter() {
+        let iter = CausalEntryIter::new(vec![]);
+        assert_eq!(iter.count(), 0);
+    }
+    
+    #[test]
+    fn test_single_queue() {
+        let node = NodeIdentity::generate();
+        let entries: VecDeque<_> = vec![
+            make_entry(&node, 1, 1000),
+            make_entry(&node, 2, 2000),
+        ].into();
+        
+        let iter = CausalEntryIter::new(vec![entries]);
+        let result: Vec<_> = iter.collect();
+        assert_eq!(result.len(), 2);
+    }
+    
+    #[test]
+    fn test_merge_multiple_queues() {
+        let node_a = NodeIdentity::generate();
+        let node_b = NodeIdentity::generate();
+        
+        // Author A: entries at time 1000, 3000
+        let queue_a: VecDeque<_> = vec![
+            make_entry(&node_a, 1, 1000),
+            make_entry(&node_a, 2, 3000),
+        ].into();
+        
+        // Author B: entries at time 2000
+        let queue_b: VecDeque<_> = vec![
+            make_entry(&node_b, 1, 2000),
+        ].into();
+        
+        let iter = CausalEntryIter::new(vec![queue_a, queue_b]);
+        let result: Vec<_> = iter.collect();
+        
+        // Should be in HLC order: 1000, 2000, 3000
+        assert_eq!(result.len(), 3);
+        
+        // Verify order by checking HLC values
+        let hlcs: Vec<_> = result.iter()
+            .map(|e| CausalEntryIter::get_hlc(e))
+            .collect();
+        assert_eq!(hlcs[0].0, 1000);
+        assert_eq!(hlcs[1].0, 2000);
+        assert_eq!(hlcs[2].0, 3000);
+    }
+    
+    #[test]
+    fn test_many_authors() {
+        // Test with 10 authors to verify heap behavior
+        let nodes: Vec<_> = (0..10).map(|_| NodeIdentity::generate()).collect();
+        let queues: Vec<VecDeque<_>> = nodes.iter().enumerate().map(|(i, node)| {
+            vec![make_entry(node, 1, (i * 100 + 50) as u64)].into()
+        }).collect();
+        
+        let iter = CausalEntryIter::new(queues);
+        let result: Vec<_> = iter.collect();
+        
+        assert_eq!(result.len(), 10);
+        
+        // Verify strictly increasing HLC order
+        let hlcs: Vec<_> = result.iter()
+            .map(|e| CausalEntryIter::get_hlc(e).0)
+            .collect();
+        for window in hlcs.windows(2) {
+            assert!(window[0] < window[1], "HLCs should be strictly increasing");
+        }
+    }
+}
+

lattice-core/src/data_dir.rs

@@ -2,19 +2,26 @@
 //!
 //! Provides platform-specific paths for Lattice data storage:
 //! - `identity.key` — Ed25519 private key
-//! - `logs/` — Append-only log files per author
+//! - `meta.db` — Global metadata (stores table)
-//! - `state.db` — KV snapshot and indexes
+//! - `stores/{uuid}/logs/{author}.log` — Per-store, per-author logs
+//! - `stores/{uuid}/state.db` — Per-store KV state
 
 use std::path::{Path, PathBuf};
+use uuid::Uuid;
 
 const APP_NAME: &str = "lattice";
 
 /// Data directory configuration.
 ///
-/// Handles paths for:
+/// Multi-store layout:
-/// - `identity.key` — node's private key
+/// ```text
-/// - `logs/{author_id}.log` — per-author log files
+/// base/
-/// - `state.db` — redb database
+///   identity.key
+///   meta.db
+///   stores/{uuid}/
+///     logs/{author}.log
+///     state.db
+/// ```
 #[derive(Debug, Clone)]
 pub struct DataDir {
     base: PathBuf,
@@ -27,10 +34,6 @@ impl DataDir {
     }
 
     /// Create a DataDir using the platform-specific data directory.
-    ///
-    /// - Linux: `~/.local/share/lattice/`
-    /// - macOS: `~/Library/Application Support/lattice/`
-    /// - Windows: `C:\Users\<user>\AppData\Roaming\lattice\`
     pub fn default_location() -> Option<Self> {
         dirs::data_dir().map(|d| Self::new(d.join(APP_NAME)))
     }
@@ -45,25 +48,47 @@ impl DataDir {
         self.base.join("identity.key")
     }
 
-    /// Get the path to the logs directory.
+    /// Get the path to the global metadata database.
-    pub fn logs_dir(&self) -> PathBuf {
+    pub fn meta_db(&self) -> PathBuf {
-        self.base.join("logs")
+        self.base.join("meta.db")
     }
 
-    /// Get the path to a specific author's log file.
+    /// Get the path to the stores directory.
-    pub fn log_file(&self, author_id_hex: &str) -> PathBuf {
+    pub fn stores_dir(&self) -> PathBuf {
-        self.logs_dir().join(format!("{}.log", author_id_hex))
+        self.base.join("stores")
     }
 
-    /// Get the path to the state database.
+    /// Get the path to a specific store's directory.
-    pub fn state_db(&self) -> PathBuf {
+    pub fn store_dir(&self, store_id: Uuid) -> PathBuf {
-        self.base.join("state.db")
+        self.stores_dir().join(store_id.to_string())
     }
 
-    /// Ensure all required directories exist.
+    /// Get the path to a store's logs directory.
+    pub fn store_logs_dir(&self, store_id: Uuid) -> PathBuf {
+        self.store_dir(store_id).join("logs")
+    }
+
+    /// Get the path to a specific author's log file within a store.
+    pub fn store_log_file(&self, store_id: Uuid, author_id_hex: &str) -> PathBuf {
+        self.store_logs_dir(store_id).join(format!("{}.log", author_id_hex))
+    }
+
+    /// Get the path to a store's state database.
+    pub fn store_state_db(&self, store_id: Uuid) -> PathBuf {
+        self.store_dir(store_id).join("state.db")
+    }
+
+    /// Ensure base directory exists.
     pub fn ensure_dirs(&self) -> std::io::Result<()> {
         std::fs::create_dir_all(&self.base)?;
-        std::fs::create_dir_all(self.logs_dir())?;
+        std::fs::create_dir_all(self.stores_dir())?;
+        Ok(())
+    }
+
+    /// Ensure directories for a specific store exist.
+    pub fn ensure_store_dirs(&self, store_id: Uuid) -> std::io::Result<()> {
+        self.ensure_dirs()?;
+        std::fs::create_dir_all(self.store_logs_dir(store_id))?;
         Ok(())
     }
 }
@@ -83,29 +108,31 @@ mod tests {
         let dd = DataDir::new("/custom/path");
         assert_eq!(dd.base(), Path::new("/custom/path"));
         assert_eq!(dd.identity_key(), PathBuf::from("/custom/path/identity.key"));
-        assert_eq!(dd.logs_dir(), PathBuf::from("/custom/path/logs"));
+        assert_eq!(dd.meta_db(), PathBuf::from("/custom/path/meta.db"));
-        assert_eq!(dd.state_db(), PathBuf::from("/custom/path/state.db"));
+        assert_eq!(dd.stores_dir(), PathBuf::from("/custom/path/stores"));
     }
 
     #[test]
-    fn test_log_file_path() {
+    fn test_store_paths() {
         let dd = DataDir::new("/data");
-        let path = dd.log_file("abc123");
+        let store_id = Uuid::parse_str("a1b2c3d4-e5f6-7890-abcd-ef1234567890").unwrap();
-        assert_eq!(path, PathBuf::from("/data/logs/abc123.log"));
+        
+        assert_eq!(dd.store_dir(store_id), PathBuf::from("/data/stores/a1b2c3d4-e5f6-7890-abcd-ef1234567890"));
+        assert_eq!(dd.store_logs_dir(store_id), PathBuf::from("/data/stores/a1b2c3d4-e5f6-7890-abcd-ef1234567890/logs"));
+        assert_eq!(dd.store_log_file(store_id, "abc123"), PathBuf::from("/data/stores/a1b2c3d4-e5f6-7890-abcd-ef1234567890/logs/abc123.log"));
+        assert_eq!(dd.store_state_db(store_id), PathBuf::from("/data/stores/a1b2c3d4-e5f6-7890-abcd-ef1234567890/state.db"));
     }
 
     #[test]
     fn test_default_location_exists() {
         // On most systems, default_location should return Some
         let location = DataDir::default_location();
-        // Just verify it doesn't panic - actual path varies by platform
         assert!(location.is_some() || true);
     }
 
     #[test]
     fn test_default_impl() {
         let dd = DataDir::default();
-        // Should either be platform default or ./data fallback
         assert!(dd.base().to_str().is_some());
     }
 }

lattice-core/src/lib.rs

@@ -1,38 +1,52 @@
 //! Lattice Core
 //!
 //! Core types for the Lattice distributed mesh:
-//! - **Node**: Identity with Ed25519 keypair
+//! - **NodeIdentity**: Cryptographic identity with Ed25519 keypair
 //! - **SigChain**: Append-only cryptographically signed log
 //! - **Entry**: Atomic operations in the log
-//! - **VectorClock**: Causality tracking for reconciliation
+//! - **SyncState**: Per-author sequence tracking for reconciliation
 //! - **HLC**: Hybrid Logical Clock for ordering
 //! - **Clock**: Time abstraction for testability
 //! - **Proto**: Generated protobuf types from lattice.proto
 //! - **DataDir**: Platform-specific data directory paths
 //! - **SignedEntry**: Entry creation, signing, and verification
 //! - **Log**: Append-only log file I/O
+//! - **Store**: Persistent KV state from log replay
+//! - **CausalIter**: Merge-sort iterator for HLC-ordered sync
 
+pub mod node_identity;
 pub mod node;
 pub mod sigchain;
 pub mod entry;
-pub mod vector_clock;
+pub mod sync_state;
 pub mod hlc;
 pub mod clock;
 pub mod proto;
 pub mod data_dir;
 pub mod signed_entry;
 pub mod log;
+pub mod store;
+pub mod meta_store;
+pub mod causal_iter;
+pub mod store_actor;
 
 // Constants
 /// Maximum size of a serialized SignedEntry (16 MB)
 pub const MAX_ENTRY_SIZE: usize = 16 * 1024 * 1024;
 
-pub use node::Node;
+pub use node_identity::{NodeIdentity, PeerStatus};
-pub use sigchain::SigChain;
+pub use node::{Node, NodeBuilder, NodeInfo, StoreInfo, StoreHandle, NodeError, NodeEvent, PeerInfo, JoinAcceptance};
+pub use sigchain::{SigChain, SigChainManager};
 pub use entry::Entry;
-pub use vector_clock::VectorClock;
+pub use sync_state::{SyncState, AuthorInfo, MissingRange};
 pub use hlc::HLC;
 pub use clock::{Clock, SystemClock, MockClock};
 pub use data_dir::DataDir;
 pub use signed_entry::{EntryBuilder, sign_entry, verify_signed_entry, hash_signed_entry};
-pub use log::{append_entry, read_entries, LogReader};
+pub use log::{append_entry, read_entries, read_entries_after, LogReader};
+pub use store::Store;
+pub use meta_store::MetaStore;
+pub use proto::HeadInfo;
+pub use uuid::Uuid;
+pub use causal_iter::CausalEntryIter;
+pub use store_actor::{StoreActor, StoreCmd, StoreActorError, spawn_store_actor};

lattice-core/src/log.rs

@@ -207,7 +207,7 @@ mod tests {
     use super::*;
     use crate::clock::MockClock;
     use crate::hlc::HLC;
-    use crate::node::Node;
+    use crate::node_identity::NodeIdentity;
     use crate::signed_entry::EntryBuilder;
     use std::env::temp_dir;
 
@@ -226,7 +226,7 @@ mod tests {
         let path = temp_log_path("single_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         let hlc = HLC::now_with_clock(&clock);
         
@@ -248,7 +248,7 @@ mod tests {
         let path = temp_log_path("multiple_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         for i in 1..=5 {
@@ -269,7 +269,7 @@ mod tests {
         let path = temp_log_path("after_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let mut entries = Vec::new();
@@ -301,7 +301,7 @@ mod tests {
         let path = temp_log_path("not_found_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
@@ -321,7 +321,7 @@ mod tests {
         let path = temp_log_path("reader_hash_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
@@ -368,7 +368,7 @@ mod tests {
         let path = temp_log_path("corrupted_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
@@ -399,7 +399,7 @@ mod tests {
         let path = temp_log_path("truncated_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
@@ -430,7 +430,7 @@ mod tests {
         let path = temp_log_path("too_large_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         // Create payload larger than MAX_ENTRY_SIZE
@@ -455,7 +455,7 @@ mod tests {
         let path = temp_log_path("boundary_last_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
@@ -506,7 +506,7 @@ mod tests {
         let path = temp_log_path("corruption_middle_v6");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         // Write 3 entries

lattice-core/src/meta_store.rs

@@ -0,0 +1,177 @@
+//! MetaStore - global node metadata in meta.db
+//!
+//! Tables:
+//! - stores: UUID → created_at (Unix ms)
+//! - meta: "root_store" → UUID (auto-opened on startup)
+
+use redb::{Database, ReadableTable, TableDefinition};
+use std::path::Path;
+use thiserror::Error;
+use uuid::Uuid;
+
+const STORES_TABLE: TableDefinition<&[u8], u64> = TableDefinition::new("stores");
+const META_TABLE: TableDefinition<&str, &[u8]> = TableDefinition::new("meta");
+
+const META_ROOT_STORE: &str = "root_store";
+const META_NAME: &str = "name";
+
+#[derive(Error, Debug)]
+pub enum MetaStoreError {
+    #[error("Database error: {0}")]
+    Database(#[from] redb::DatabaseError),
+    
+    #[error("Table error: {0}")]
+    Table(#[from] redb::TableError),
+    
+    #[error("Transaction error: {0}")]
+    Transaction(#[from] redb::TransactionError),
+    
+    #[error("Commit error: {0}")]
+    Commit(#[from] redb::CommitError),
+    
+    #[error("Storage error: {0}")]
+    Storage(#[from] redb::StorageError),
+}
+
+/// Global metadata store
+pub struct MetaStore {
+    db: Database,
+}
+
+impl MetaStore {
+    /// Open or create meta.db at the given path
+    pub fn open(path: impl AsRef<Path>) -> Result<Self, MetaStoreError> {
+        let db = Database::create(path)?;
+        
+        // Ensure tables exist
+        let write_txn = db.begin_write()?;
+        {
+            let _ = write_txn.open_table(STORES_TABLE)?;
+            let _ = write_txn.open_table(META_TABLE)?;
+        }
+        write_txn.commit()?;
+        
+        Ok(Self { db })
+    }
+
+    /// Register a new store
+    pub fn add_store(&self, store_id: Uuid) -> Result<(), MetaStoreError> {
+        let write_txn = self.db.begin_write()?;
+        {
+            let mut table = write_txn.open_table(STORES_TABLE)?;
+            let now = std::time::SystemTime::now()
+                .duration_since(std::time::UNIX_EPOCH)
+                .unwrap()
+                .as_millis() as u64;
+            table.insert(store_id.as_bytes().as_slice(), now)?;
+        }
+        write_txn.commit()?;
+        Ok(())
+    }
+
+    /// List all registered stores
+    pub fn list_stores(&self) -> Result<Vec<Uuid>, MetaStoreError> {
+        let read_txn = self.db.begin_read()?;
+        let table = read_txn.open_table(STORES_TABLE)?;
+        
+        let mut stores = Vec::new();
+        for result in table.iter()? {
+            let (key, _created_at) = result?;
+            let bytes: [u8; 16] = key.value().try_into().unwrap_or([0; 16]);
+            stores.push(Uuid::from_bytes(bytes));
+        }
+        Ok(stores)
+    }
+
+    /// Get the root store ID (auto-opened on startup)
+    pub fn root_store(&self) -> Result<Option<Uuid>, MetaStoreError> {
+        let read_txn = self.db.begin_read()?;
+        let table = read_txn.open_table(META_TABLE)?;
+        
+        match table.get(META_ROOT_STORE)? {
+            Some(value) => {
+                let bytes: [u8; 16] = value.value().try_into().unwrap_or([0; 16]);
+                Ok(Some(Uuid::from_bytes(bytes)))
+            }
+            None => Ok(None),
+        }
+    }
+
+    /// Set the root store ID
+    pub fn set_root_store(&self, store_id: Uuid) -> Result<(), MetaStoreError> {
+        let write_txn = self.db.begin_write()?;
+        {
+            let mut table = write_txn.open_table(META_TABLE)?;
+            table.insert(META_ROOT_STORE, store_id.as_bytes().as_slice())?;
+        }
+        write_txn.commit()?;
+        Ok(())
+    }
+    
+    /// Get the node's display name
+    pub fn name(&self) -> Result<Option<String>, MetaStoreError> {
+        let read_txn = self.db.begin_read()?;
+        let table = read_txn.open_table(META_TABLE)?;
+        
+        match table.get(META_NAME)? {
+            Some(value) => Ok(Some(String::from_utf8_lossy(value.value()).to_string())),
+            None => Ok(None),
+        }
+    }
+    
+    /// Set the node's display name
+    pub fn set_name(&self, name: &str) -> Result<(), MetaStoreError> {
+        let write_txn = self.db.begin_write()?;
+        {
+            let mut table = write_txn.open_table(META_TABLE)?;
+            table.insert(META_NAME, name.as_bytes())?;
+        }
+        write_txn.commit()?;
+        Ok(())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::env::temp_dir;
+
+    #[test]
+    fn test_add_and_list_stores() {
+        let path = temp_dir().join("meta_store_test.db");
+        let _ = std::fs::remove_file(&path);
+        
+        let meta = MetaStore::open(&path).unwrap();
+        
+        let id1 = Uuid::new_v4();
+        let id2 = Uuid::new_v4();
+        
+        meta.add_store(id1).unwrap();
+        meta.add_store(id2).unwrap();
+        
+        let stores = meta.list_stores().unwrap();
+        assert_eq!(stores.len(), 2);
+        assert!(stores.contains(&id1));
+        assert!(stores.contains(&id2));
+        
+        let _ = std::fs::remove_file(&path);
+    }
+
+    #[test]
+    fn test_root_store() {
+        let path = temp_dir().join("meta_store_root.db");
+        let _ = std::fs::remove_file(&path);
+        
+        let meta = MetaStore::open(&path).unwrap();
+        
+        // Initially no root store
+        assert_eq!(meta.root_store().unwrap(), None);
+        
+        let root = Uuid::new_v4();
+        meta.set_root_store(root).unwrap();
+        
+        assert_eq!(meta.root_store().unwrap(), Some(root));
+        
+        let _ = std::fs::remove_file(&path);
+    }
+}

lattice-core/src/node_identity.rs

@@ -0,0 +1,252 @@
+//! Node identity and cryptographic keys
+//!
+//! Each node has an Ed25519 keypair:
+//! - Private key: stored locally in `identity.key` (never replicated)
+//! - Public key: serves as the node's identity (32 bytes)
+
+use ed25519_dalek::{Signature, Signer, SigningKey, Verifier, VerifyingKey};
+use rand::rngs::OsRng;
+use std::fs;
+use std::io::{self, Read, Write};
+use std::path::Path;
+use thiserror::Error;
+
+/// Errors that can occur during node operations
+#[derive(Error, Debug)]
+pub enum NodeError {
+    #[error("IO error: {0}")]
+    Io(#[from] io::Error),
+    
+    #[error("Invalid key length: expected 32 bytes, got {0}")]
+    InvalidKeyLength(usize),
+    
+    #[error("Invalid signature")]
+    InvalidSignature,
+}
+
+/// A node in the Lattice mesh.
+///
+/// Each node has an Ed25519 keypair used for signing sigchain entries
+/// and establishing trust within the network.
+#[derive(Clone)]
+pub struct NodeIdentity {
+    signing_key: SigningKey,
+}
+
+impl NodeIdentity {
+    /// Generate a new node with a random keypair.
+    pub fn generate() -> Self {
+        let signing_key = SigningKey::generate(&mut OsRng);
+        Self { signing_key }
+    }
+
+    /// Create a node from an existing signing key.
+    pub fn from_signing_key(signing_key: SigningKey) -> Self {
+        Self { signing_key }
+    }
+
+    /// Load a node's identity from a key file, or generate and save if it doesn't exist.
+    pub fn load_or_generate(path: impl AsRef<Path>) -> Result<Self, NodeError> {
+        let path = path.as_ref();
+        if path.exists() {
+            Self::load(path)
+        } else {
+            let node = Self::generate();
+            node.save(path)?;
+            Ok(node)
+        }
+    }
+
+    /// Load a node's identity from a key file.
+    pub fn load(path: impl AsRef<Path>) -> Result<Self, NodeError> {
+        let mut file = fs::File::open(path)?;
+        let mut bytes = Vec::new();
+        file.read_to_end(&mut bytes)?;
+        
+        if bytes.len() != 32 {
+            return Err(NodeError::InvalidKeyLength(bytes.len()));
+        }
+        
+        let key_bytes: [u8; 32] = bytes.try_into().unwrap();
+        let signing_key = SigningKey::from_bytes(&key_bytes);
+        Ok(Self { signing_key })
+    }
+
+    /// Save the node's private key to a file.
+    pub fn save(&self, path: impl AsRef<Path>) -> Result<(), NodeError> {
+        let path = path.as_ref();
+        
+        // Create parent directories if they don't exist
+        if let Some(parent) = path.parent() {
+            fs::create_dir_all(parent)?;
+        }
+        
+        let mut file = fs::File::create(path)?;
+        file.write_all(self.signing_key.as_bytes())?;
+        Ok(())
+    }
+
+    /// Get the node's public key (identity).
+    pub fn public_key(&self) -> VerifyingKey {
+        self.signing_key.verifying_key()
+    }
+
+    /// Get the node's public key as bytes (32 bytes).
+    pub fn public_key_bytes(&self) -> [u8; 32] {
+        self.signing_key.verifying_key().to_bytes()
+    }
+
+    /// Get the signing key for creating signatures.
+    pub fn signing_key(&self) -> &SigningKey {
+        &self.signing_key
+    }
+
+    /// Get the secret key bytes (32 bytes) for Iroh integration.
+    /// WARNING: Handle with care - this exposes the private key material.
+    pub fn secret_key_bytes(&self) -> [u8; 32] {
+        self.signing_key.to_bytes()
+    }
+
+    /// Sign a message.
+    pub fn sign(&self, message: &[u8]) -> Signature {
+        self.signing_key.sign(message)
+    }
+
+    /// Verify a signature against this node's public key.
+    pub fn verify(&self, message: &[u8], signature: &Signature) -> Result<(), NodeError> {
+        self.public_key()
+            .verify(message, signature)
+            .map_err(|_| NodeError::InvalidSignature)
+    }
+
+    /// Verify a signature using a raw public key.
+    pub fn verify_with_key(
+        public_key: &VerifyingKey,
+        message: &[u8],
+        signature: &Signature,
+    ) -> Result<(), NodeError> {
+        public_key
+            .verify(message, signature)
+            .map_err(|_| NodeError::InvalidSignature)
+    }
+}
+
+/// Peer status values used across the system
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum PeerStatus {
+    /// Peer has been invited but hasn't joined yet
+    Invited,
+    /// Peer is active and can sync
+    Active,
+    /// Peer is temporarily inactive
+    Dormant,
+}
+
+impl PeerStatus {
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            PeerStatus::Invited => "invited",
+            PeerStatus::Active => "active",
+            PeerStatus::Dormant => "dormant",
+        }
+    }
+    
+    pub fn from_str(s: &str) -> Option<PeerStatus> {
+        match s {
+            "invited" => Some(PeerStatus::Invited),
+            "active" => Some(PeerStatus::Active),
+            "dormant" => Some(PeerStatus::Dormant),
+            _ => None,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::env::temp_dir;
+
+    #[test]
+    fn test_generate() {
+        let node = NodeIdentity::generate();
+        let pk = node.public_key_bytes();
+        assert_eq!(pk.len(), 32);
+    }
+
+    #[test]
+    fn test_sign_and_verify() {
+        let node = NodeIdentity::generate();
+        let message = b"hello lattice";
+        
+        let signature = node.sign(message);
+        assert!(node.verify(message, &signature).is_ok());
+    }
+
+    #[test]
+    fn test_verify_wrong_message() {
+        let node = NodeIdentity::generate();
+        let signature = node.sign(b"original");
+        
+        assert!(node.verify(b"tampered", &signature).is_err());
+    }
+
+    #[test]
+    fn test_verify_with_different_key() {
+        let node1 = NodeIdentity::generate();
+        let node2 = NodeIdentity::generate();
+        
+        let signature = node1.sign(b"message");
+        assert!(node2.verify(b"message", &signature).is_err());
+    }
+
+    #[test]
+    fn test_save_and_load() {
+        let temp_path = temp_dir().join("lattice_test_identity.key");
+        
+        // Generate and save
+        let node1 = NodeIdentity::generate();
+        let pk1 = node1.public_key_bytes();
+        node1.save(&temp_path).unwrap();
+        
+        // Load and verify same key
+        let node2 = NodeIdentity::load(&temp_path).unwrap();
+        let pk2 = node2.public_key_bytes();
+        
+        assert_eq!(pk1, pk2);
+        
+        // Cleanup
+        fs::remove_file(&temp_path).ok();
+    }
+
+    #[test]
+    fn test_load_or_generate() {
+        let temp_path = temp_dir().join("lattice_test_identity2.key");
+        
+        // Remove if exists
+        fs::remove_file(&temp_path).ok();
+        
+        // First call: generates
+        let node1 = NodeIdentity::load_or_generate(&temp_path).unwrap();
+        let pk1 = node1.public_key_bytes();
+        
+        // Second call: loads existing
+        let node2 = NodeIdentity::load_or_generate(&temp_path).unwrap();
+        let pk2 = node2.public_key_bytes();
+        
+        assert_eq!(pk1, pk2);
+        
+        // Cleanup
+        fs::remove_file(&temp_path).ok();
+    }
+
+    #[test]
+    fn test_verify_with_key_static() {
+        let node = NodeIdentity::generate();
+        let pk = node.public_key();
+        let message = b"test message";
+        
+        let signature = node.sign(message);
+        
+        assert!(NodeIdentity::verify_with_key(&pk, message, &signature).is_ok());
+    }
+}

lattice-core/src/proto.rs

@@ -31,7 +31,9 @@ mod tests {
     fn test_entry_with_ops() {
         let entry = Entry {
             version: 1,
+            store_id: vec![1u8; 16],
             prev_hash: vec![0u8; 32],
+            parent_hashes: vec![],
             seq: 5,
             timestamp: Some(Hlc {
                 wall_time: 1000,
@@ -40,7 +42,7 @@ mod tests {
             ops: vec![
                 Operation {
                     op_type: Some(operation::OpType::Put(PutOp {
-                        key: "/nodes/abc".to_string(),
+                        key: b"/nodes/abc".to_vec(),
                         value: b"hello".to_vec(),
                     })),
                 },

lattice-core/src/sigchain.rs

@@ -4,7 +4,7 @@
 //! before appending (correct seq, prev_hash, valid signature) and persists to disk.
 
 use crate::log::{append_entry, read_entries, LogError};
-use crate::node::Node;
+use crate::node_identity::NodeIdentity;
 use crate::proto::{Entry, SignedEntry};
 use crate::signed_entry::{hash_signed_entry, verify_signed_entry};
 use prost::Message;
@@ -23,6 +23,9 @@ pub enum SigChainError {
     #[error("Wrong author: expected {expected}, got {got}")]
     WrongAuthor { expected: String, got: String },
     
+    #[error("Wrong store_id: expected {expected}, got {got}")]
+    WrongStoreId { expected: String, got: String },
+    
     #[error("Invalid sequence: expected {expected}, got {got}")]
     InvalidSequence { expected: u64, got: u64 },
     
@@ -34,11 +37,14 @@ pub enum SigChainError {
 }
 
 /// An append-only log where each entry is cryptographically signed
-/// and hash-linked to the previous entry.
+/// and hash-linked to the previous entry, scoped to a specific store.
 pub struct SigChain {
     /// Path to the log file
     log_path: PathBuf,
     
+    /// Store UUID (16 bytes)
+    store_id: [u8; 16],
+    
     /// Author's public key (32 bytes)
     author_id: [u8; 32],
     
@@ -50,10 +56,11 @@ pub struct SigChain {
 }
 
 impl SigChain {
-    /// Create a new empty sigchain for an author
+    /// Create a new empty sigchain for a (store, author) pair
-    pub fn new(log_path: impl AsRef<Path>, author_id: [u8; 32]) -> Self {
+    pub fn new(log_path: impl AsRef<Path>, store_id: [u8; 16], author_id: [u8; 32]) -> Self {
         Self {
             log_path: log_path.as_ref().to_path_buf(),
+            store_id,
             author_id,
             next_seq: 1,
             last_hash: [0u8; 32],
@@ -61,11 +68,11 @@ impl SigChain {
     }
     
     /// Load a sigchain from an existing log file
-    pub fn from_log(log_path: impl AsRef<Path>, author_id: [u8; 32]) -> Result<Self, SigChainError> {
+    pub fn from_log(log_path: impl AsRef<Path>, store_id: [u8; 16], author_id: [u8; 32]) -> Result<Self, SigChainError> {
         let log_path = log_path.as_ref().to_path_buf();
         let entries = read_entries(&log_path)?;
         
-        let mut chain = Self::new(&log_path, author_id);
+        let mut chain = Self::new(&log_path, store_id, author_id);
         
         for signed_entry in entries {
             // Verify signature
@@ -85,6 +92,18 @@ impl SigChain {
             // Decode Entry
             let entry = Entry::decode(&signed_entry.entry_bytes[..])?;
             
+            // Validate store_id
+            // Note: Empty/malformed store_id becomes [0u8;16], which fails validation
+            // against any real UUID store. This intentionally rejects legacy entries.
+            let entry_store: [u8; 16] = entry.store_id.clone().try_into()
+                .unwrap_or([0u8; 16]);
+            if entry_store != store_id {
+                return Err(SigChainError::WrongStoreId {
+                    expected: hex::encode(store_id),
+                    got: hex::encode(entry_store),
+                });
+            }
+            
             // Validate sequence
             if entry.seq != chain.next_seq {
                 return Err(SigChainError::InvalidSequence {
@@ -127,6 +146,11 @@ impl SigChain {
         &self.last_hash
     }
     
+    /// Get the log file path
+    pub fn log_path(&self) -> &std::path::Path {
+        &self.log_path
+    }
+    
     /// Get the current length of the chain
     pub fn len(&self) -> u64 {
         self.next_seq - 1
@@ -156,6 +180,18 @@ impl SigChain {
         // Decode entry
         let entry = Entry::decode(&signed_entry.entry_bytes[..])?;
         
+        // Validate store_id
+        // Note: Empty/malformed store_id becomes [0u8;16], which fails validation
+        // against any real UUID store. This intentionally rejects legacy entries.
+        let entry_store: [u8; 16] = entry.store_id.clone().try_into()
+            .unwrap_or([0u8; 16]);
+        if entry_store != self.store_id {
+            return Err(SigChainError::WrongStoreId {
+                expected: hex::encode(self.store_id),
+                got: hex::encode(entry_store),
+            });
+        }
+        
         // Validate sequence
         if entry.seq != self.next_seq {
             return Err(SigChainError::InvalidSequence {
@@ -193,7 +229,7 @@ impl SigChain {
     }
     
     /// Create and append a new entry using the node's key
-    pub fn create_entry(&mut self, node: &Node, ops: Vec<crate::proto::Operation>) -> Result<SignedEntry, SigChainError> {
+    pub fn create_entry(&mut self, node: &NodeIdentity, ops: Vec<crate::proto::Operation>) -> Result<SignedEntry, SigChainError> {
         use crate::clock::SystemClock;
         use crate::hlc::HLC;
         use crate::signed_entry::EntryBuilder;
@@ -201,6 +237,7 @@ impl SigChain {
         let hlc = HLC::now_with_clock(&SystemClock);
         
         let mut builder = EntryBuilder::new(self.next_seq, hlc)
+            .store_id(self.store_id.to_vec())
             .prev_hash(self.last_hash.to_vec());
         
         // Add operations
@@ -216,12 +253,96 @@ impl SigChain {
     }
 }
 
+/// Manages multiple SigChains (one per author) for a store.
+/// Provides unified interface for appending entries from any author.
+pub struct SigChainManager {
+    /// Directory containing log files (one per author)
+    logs_dir: PathBuf,
+    
+    /// Store UUID (16 bytes)
+    store_id: [u8; 16],
+    
+    /// Cache of loaded SigChains by author
+    chains: std::collections::HashMap<[u8; 32], SigChain>,
+}
+
+impl SigChainManager {
+    /// Create a new manager for a store's logs directory
+    pub fn new(logs_dir: impl AsRef<Path>, store_id: [u8; 16]) -> Self {
+        Self {
+            logs_dir: logs_dir.as_ref().to_path_buf(),
+            store_id,
+            chains: std::collections::HashMap::new(),
+        }
+    }
+    
+    /// Get or create a SigChain for an author
+    pub fn get_or_create(&mut self, author: [u8; 32]) -> &mut SigChain {
+        self.chains.entry(author).or_insert_with(|| {
+            let author_hex = hex::encode(author);
+            let log_path = self.logs_dir.join(format!("{}.log", author_hex));
+            
+            // Try to load existing log, or create new
+            SigChain::from_log(&log_path, self.store_id, author)
+                .unwrap_or_else(|_| SigChain::new(&log_path, self.store_id, author))
+        })
+    }
+    
+    /// Get the local node's sigchain (for creating new entries)
+    pub fn get(&self, author: &[u8; 32]) -> Option<&SigChain> {
+        self.chains.get(author)
+    }
+    
+    /// Append an entry to the appropriate author's log
+    /// This is the main entry point for all entry writes (from put, sync, etc.)
+    pub fn append_entry(&mut self, entry: &SignedEntry) -> Result<(), SigChainError> {
+        let author: [u8; 32] = entry.author_id.clone()
+            .try_into()
+            .map_err(|_| SigChainError::WrongAuthor {
+                expected: "32 bytes".to_string(),
+                got: format!("{} bytes", entry.author_id.len()),
+            })?;
+        
+        // For synced entries, we can't validate seq/prev_hash since they may arrive
+        // out of order. Just append to the log file directly.
+        let chain = self.get_or_create(author);
+        append_entry(chain.log_path(), entry)?;
+        
+        Ok(())
+    }
+    
+    /// Get the logs directory path
+    pub fn logs_dir(&self) -> &Path {
+        &self.logs_dir
+    }
+    
+    /// Get log directory statistics (file count, total bytes)
+    pub fn log_stats(&self) -> (usize, u64) {
+        if !self.logs_dir.exists() {
+            return (0, 0);
+        }
+        let mut total_size = 0u64;
+        let mut file_count = 0;
+        if let Ok(entries) = std::fs::read_dir(&self.logs_dir) {
+            for entry in entries.flatten() {
+                if let Ok(meta) = entry.metadata() {
+                    if meta.is_file() {
+                        total_size += meta.len();
+                        file_count += 1;
+                    }
+                }
+            }
+        }
+        (file_count, total_size)
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
     use crate::clock::MockClock;
     use crate::hlc::HLC;
-    use crate::node::Node;
+    use crate::node_identity::NodeIdentity;
     use crate::proto::{operation, Operation, PutOp};
     use crate::signed_entry::EntryBuilder;
     use std::env::temp_dir;
@@ -230,12 +351,14 @@ mod tests {
         temp_dir().join(format!("lattice_sigchain_test_{}.log", name))
     }
 
+    const TEST_STORE: [u8; 16] = [1u8; 16];
+
     #[test]
     fn test_new_sigchain() {
         let path = temp_log_path("new");
         let author = [1u8; 32];
         
-        let chain = SigChain::new(&path, author);
+        let chain = SigChain::new(&path, TEST_STORE, author);
         
         assert_eq!(chain.author_id(), &author);
         assert_eq!(chain.next_seq(), 1);
@@ -249,12 +372,13 @@ mod tests {
         let path = temp_log_path("append");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
-        let mut chain = SigChain::new(&path, author);
+        let mut chain = SigChain::new(&path, TEST_STORE, author);
         
         let clock = MockClock::new(1000);
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
+            .store_id(TEST_STORE.to_vec())
             .prev_hash([0u8; 32].to_vec())
             .put("/key", b"value".to_vec())
             .sign(&node);
@@ -273,13 +397,14 @@ mod tests {
         let path = temp_log_path("multiple");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
-        let mut chain = SigChain::new(&path, author);
+        let mut chain = SigChain::new(&path, TEST_STORE, author);
         let clock = MockClock::new(1000);
         
         for i in 1..=3 {
             let entry = EntryBuilder::new(i, HLC::now_with_clock(&clock))
+                .store_id(TEST_STORE.to_vec())
                 .prev_hash(chain.last_hash.to_vec())
                 .put(format!("/key/{}", i), format!("value{}", i).into_bytes())
                 .sign(&node);
@@ -297,15 +422,16 @@ mod tests {
         let path = temp_log_path("from_log");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
         let clock = MockClock::new(1000);
         
         // Write some entries
         {
-            let mut chain = SigChain::new(&path, author);
+            let mut chain = SigChain::new(&path, TEST_STORE, author);
             for i in 1..=3 {
                 let entry = EntryBuilder::new(i, HLC::now_with_clock(&clock))
+                    .store_id(TEST_STORE.to_vec())
                     .prev_hash(chain.last_hash.to_vec())
                     .put("/key", b"val".to_vec())
                     .sign(&node);
@@ -314,7 +440,7 @@ mod tests {
         }
         
         // Reload from log
-        let chain = SigChain::from_log(&path, author).unwrap();
+        let chain = SigChain::from_log(&path, TEST_STORE, author).unwrap();
         
         assert_eq!(chain.len(), 3);
         assert_eq!(chain.next_seq(), 4);
@@ -327,13 +453,14 @@ mod tests {
         let path = temp_log_path("wrong_seq");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
-        let mut chain = SigChain::new(&path, author);
+        let mut chain = SigChain::new(&path, TEST_STORE, author);
         let clock = MockClock::new(1000);
         
         // Try to append with wrong seq (2 instead of 1)
         let entry = EntryBuilder::new(2, HLC::now_with_clock(&clock))
+            .store_id(TEST_STORE.to_vec())
             .prev_hash([0u8; 32].to_vec())
             .put("/key", b"val".to_vec())
             .sign(&node);
@@ -350,13 +477,14 @@ mod tests {
         let path = temp_log_path("wrong_prev");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
-        let mut chain = SigChain::new(&path, author);
+        let mut chain = SigChain::new(&path, TEST_STORE, author);
         let clock = MockClock::new(1000);
         
         // First entry
         let entry1 = EntryBuilder::new(1, HLC::now_with_clock(&clock))
+            .store_id(TEST_STORE.to_vec())
             .prev_hash([0u8; 32].to_vec())
             .put("/key", b"v1".to_vec())
             .sign(&node);
@@ -364,6 +492,7 @@ mod tests {
         
         // Second entry with wrong prev_hash
         let entry2 = EntryBuilder::new(2, HLC::now_with_clock(&clock))
+            .store_id(TEST_STORE.to_vec())
             .prev_hash([99u8; 32].to_vec()) // Wrong!
             .put("/key", b"v2".to_vec())
             .sign(&node);
@@ -380,13 +509,14 @@ mod tests {
         let path = temp_log_path("wrong_author");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let other_author = [99u8; 32]; // Different author
-        let mut chain = SigChain::new(&path, other_author);
+        let mut chain = SigChain::new(&path, TEST_STORE, other_author);
         let clock = MockClock::new(1000);
         
         // Entry signed by node but chain expects other_author
         let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
+            .store_id(TEST_STORE.to_vec())
             .prev_hash([0u8; 32].to_vec())
             .put("/key", b"val".to_vec())
             .sign(&node);
@@ -403,14 +533,14 @@ mod tests {
         let path = temp_log_path("create");
         std::fs::remove_file(&path).ok();
         
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let author = node.public_key_bytes();
-        let mut chain = SigChain::new(&path, author);
+        let mut chain = SigChain::new(&path, TEST_STORE, author);
         
         let ops = vec![
             Operation {
                 op_type: Some(operation::OpType::Put(PutOp {
-                    key: "/test".to_string(),
+                    key: b"/test".to_vec(),
                     value: b"hello".to_vec(),
                 })),
             },
@@ -427,4 +557,37 @@ mod tests {
         
         std::fs::remove_file(&path).ok();
     }
+
+    #[test]
+    fn test_reject_wrong_store_id() {
+        let path_a = temp_log_path("storeid_a");
+        let path_b = temp_log_path("storeid_b");
+        std::fs::remove_file(&path_a).ok();
+        std::fs::remove_file(&path_b).ok();
+        
+        let node = NodeIdentity::generate();
+        let author = node.public_key_bytes();
+        let clock = MockClock::new(1000);
+        
+        let store_a = [0xAAu8; 16];
+        let store_b = [0xBBu8; 16];
+        
+        // Create valid entry for store A
+        let mut chain_a = SigChain::new(&path_a, store_a, author);
+        let entry = EntryBuilder::new(1, HLC::now_with_clock(&clock))
+            .store_id(store_a.to_vec())
+            .prev_hash([0u8; 32].to_vec())
+            .put("/key", b"val".to_vec())
+            .sign(&node);
+        chain_a.append(&entry).unwrap();
+        
+        // Try to replay that entry into store B's chain
+        let mut chain_b = SigChain::new(&path_b, store_b, author);
+        let result = chain_b.append(&entry);
+        
+        assert!(matches!(result, Err(SigChainError::WrongStoreId { .. })));
+        
+        std::fs::remove_file(&path_a).ok();
+        std::fs::remove_file(&path_b).ok();
+    }
 }

lattice-core/src/signed_entry.rs

@@ -7,7 +7,7 @@
 //! - Computing entry hashes for prev_hash linking
 
 use crate::hlc::HLC;
-use crate::node::{Node, NodeError};
+use crate::node_identity::{NodeIdentity, NodeError};
 use crate::proto::{Entry, Hlc, Operation, PutOp, DeleteOp, SignedEntry, operation};
 use ed25519_dalek::{Signature, VerifyingKey};
 use prost::Message;
@@ -32,7 +32,9 @@ pub enum EntryError {
 /// Builder for creating Entry messages
 pub struct EntryBuilder {
     version: u32,
+    store_id: Vec<u8>,
     prev_hash: Vec<u8>,
+    parent_hashes: Vec<Vec<u8>>,
     seq: u64,
     timestamp: HLC,
     ops: Vec<Operation>,
@@ -43,21 +45,35 @@ impl EntryBuilder {
     pub fn new(seq: u64, timestamp: HLC) -> Self {
         Self {
             version: 1,
-            prev_hash: vec![0u8; 32], // Genesis or will be set
+            store_id: Vec::new(),
+            prev_hash: vec![0u8; 32],
+            parent_hashes: Vec::new(),
             seq,
             timestamp,
             ops: Vec::new(),
         }
     }
 
-    /// Set the previous entry hash (for chaining)
+    /// Set the store ID (16-byte UUID)
+    pub fn store_id(mut self, id: impl Into<Vec<u8>>) -> Self {
+        self.store_id = id.into();
+        self
+    }
+
+    /// Set the previous entry hash (for sigchain linking)
     pub fn prev_hash(mut self, hash: impl Into<Vec<u8>>) -> Self {
         self.prev_hash = hash.into();
         self
     }
 
+    /// Set the parent hashes (for DAG ancestry)
+    pub fn parent_hashes(mut self, hashes: Vec<Vec<u8>>) -> Self {
+        self.parent_hashes = hashes;
+        self
+    }
+
     /// Add a Put operation
-    pub fn put(mut self, key: impl Into<String>, value: impl Into<Vec<u8>>) -> Self {
+    pub fn put(mut self, key: impl Into<Vec<u8>>, value: impl Into<Vec<u8>>) -> Self {
         self.ops.push(Operation {
             op_type: Some(operation::OpType::Put(PutOp {
                 key: key.into(),
@@ -68,7 +84,7 @@ impl EntryBuilder {
     }
 
     /// Add a Delete operation
-    pub fn delete(mut self, key: impl Into<String>) -> Self {
+    pub fn delete(mut self, key: impl Into<Vec<u8>>) -> Self {
         self.ops.push(Operation {
             op_type: Some(operation::OpType::Delete(DeleteOp {
                 key: key.into(),
@@ -87,7 +103,9 @@ impl EntryBuilder {
     pub fn build(self) -> Entry {
         Entry {
             version: self.version,
+            store_id: self.store_id,
             prev_hash: self.prev_hash,
+            parent_hashes: self.parent_hashes,
             seq: self.seq,
             timestamp: Some(Hlc {
                 wall_time: self.timestamp.wall_time,
@@ -98,14 +116,14 @@ impl EntryBuilder {
     }
 
     /// Build and sign the entry, returning a SignedEntry
-    pub fn sign(self, node: &Node) -> SignedEntry {
+    pub fn sign(self, node: &NodeIdentity) -> SignedEntry {
         let entry = self.build();
         sign_entry(&entry, node)
     }
 }
 
 /// Sign an Entry to create a SignedEntry
-pub fn sign_entry(entry: &Entry, node: &Node) -> SignedEntry {
+pub fn sign_entry(entry: &Entry, node: &NodeIdentity) -> SignedEntry {
     let entry_bytes = entry.encode_to_vec();
     let signature = node.sign(&entry_bytes);
     
@@ -134,7 +152,7 @@ pub fn verify_signed_entry(signed: &SignedEntry) -> Result<Entry, EntryError> {
     let signature = Signature::from_bytes(&sig_bytes);
     
     // Verify
-    Node::verify_with_key(&public_key, &signed.entry_bytes, &signature)?;
+    NodeIdentity::verify_with_key(&public_key, &signed.entry_bytes, &signature)?;
     
     // Decode entry
     let entry = Entry::decode(&signed.entry_bytes[..])?;
@@ -174,7 +192,7 @@ mod tests {
 
     #[test]
     fn test_sign_and_verify() {
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         let hlc = HLC::now_with_clock(&clock);
         
@@ -193,7 +211,7 @@ mod tests {
 
     #[test]
     fn test_verify_tampered_fails() {
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         let hlc = HLC::now_with_clock(&clock);
         
@@ -209,8 +227,8 @@ mod tests {
 
     #[test]
     fn test_verify_wrong_key_fails() {
-        let node1 = Node::generate();
+        let node1 = NodeIdentity::generate();
-        let node2 = Node::generate();
+        let node2 = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         let hlc = HLC::now_with_clock(&clock);
         
@@ -226,7 +244,7 @@ mod tests {
 
     #[test]
     fn test_hash_signed_entry() {
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         let hlc = HLC::now_with_clock(&clock);
         
@@ -244,7 +262,7 @@ mod tests {
 
     #[test]
     fn test_prev_hash_chaining() {
-        let node = Node::generate();
+        let node = NodeIdentity::generate();
         let clock = MockClock::new(1000);
         
         // First entry

lattice-core/src/store_actor.rs

@@ -0,0 +1,299 @@
+//! Store Actor - dedicated thread that owns Store and processes commands via channel
+
+use crate::{
+    EntryBuilder, HeadInfo, NodeIdentity, SigChain, SigChainManager, Store, Uuid,
+    hlc::HLC,
+    proto::AuthorState,
+    sigchain::SigChainError,
+    store::StoreError,
+    sync_state::SyncState,
+    proto::SignedEntry,
+    log,
+};
+use tokio::sync::{mpsc, oneshot, broadcast};
+use std::thread::{self, JoinHandle};
+
+/// Commands sent to the store actor
+pub enum StoreCmd {
+    Get {
+        key: Vec<u8>,
+        resp: oneshot::Sender<Result<Option<Vec<u8>>, StoreError>>,
+    },
+    GetHeads {
+        key: Vec<u8>,
+        resp: oneshot::Sender<Result<Vec<HeadInfo>, StoreError>>,
+    },
+    List {
+        include_deleted: bool,
+        resp: oneshot::Sender<Result<Vec<(Vec<u8>, Vec<u8>)>, StoreError>>,
+    },
+    ListByPrefix {
+        prefix: Vec<u8>,
+        include_deleted: bool,
+        resp: oneshot::Sender<Result<Vec<(Vec<u8>, Vec<u8>)>, StoreError>>,
+    },
+    Put {
+        key: Vec<u8>,
+        value: Vec<u8>,
+        resp: oneshot::Sender<Result<u64, StoreActorError>>,
+    },
+    Delete {
+        key: Vec<u8>,
+        resp: oneshot::Sender<Result<u64, StoreActorError>>,
+    },
+    LogSeq {
+        resp: oneshot::Sender<u64>,
+    },
+    AppliedSeq {
+        resp: oneshot::Sender<Result<u64, StoreError>>,
+    },
+    AuthorState {
+        author: [u8; 32],
+        resp: oneshot::Sender<Result<Option<AuthorState>, StoreError>>,
+    },
+    // Sync-related commands
+    SyncState {
+        resp: oneshot::Sender<Result<SyncState, StoreError>>,
+    },
+    ReadEntriesAfter {
+        author: [u8; 32],
+        from_hash: Option<[u8; 32]>,
+        resp: oneshot::Sender<Result<Vec<SignedEntry>, StoreError>>,
+    },
+    ApplyEntry {
+        entry: SignedEntry,
+        resp: oneshot::Sender<Result<(), StoreError>>,
+    },
+    LogStats {
+        resp: oneshot::Sender<(usize, u64)>,
+    },
+    Shutdown,
+}
+
+#[derive(Debug)]
+pub enum StoreActorError {
+    Store(StoreError),
+    SigChain(SigChainError),
+}
+
+impl From<StoreError> for StoreActorError {
+    fn from(e: StoreError) -> Self {
+        StoreActorError::Store(e)
+    }
+}
+
+impl From<SigChainError> for StoreActorError {
+    fn from(e: SigChainError) -> Self {
+        StoreActorError::SigChain(e)
+    }
+}
+
+impl std::fmt::Display for StoreActorError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            StoreActorError::Store(e) => write!(f, "Store error: {}", e),
+            StoreActorError::SigChain(e) => write!(f, "SigChain error: {}", e),
+        }
+    }
+}
+
+impl std::error::Error for StoreActorError {}
+
+/// The store actor - runs in its own thread, owns Store and SigChainManager
+pub struct StoreActor {
+    store_id: Uuid,
+    store: Store,
+    chain_manager: SigChainManager,
+    node: NodeIdentity,
+    rx: mpsc::Receiver<StoreCmd>,
+    /// Broadcast sender for emitting entries after they're committed locally
+    entry_tx: broadcast::Sender<SignedEntry>,
+}
+
+impl StoreActor {
+    /// Create a new store actor (but don't start the thread yet)
+    pub fn new(
+        store_id: Uuid,
+        store: Store,
+        sigchain: SigChain,
+        node: NodeIdentity,
+        rx: mpsc::Receiver<StoreCmd>,
+        entry_tx: broadcast::Sender<SignedEntry>,
+    ) -> Self {
+        // Derive logs_dir from sigchain's log file path
+        let logs_dir = sigchain.log_path()
+            .parent()
+            .map(|p| p.to_path_buf())
+            .unwrap_or_default();
+        
+        // Create chain manager and register the local node's sigchain
+        let mut chain_manager = SigChainManager::new(&logs_dir, *store_id.as_bytes());
+        let local_author = node.public_key_bytes();
+        chain_manager.get_or_create(local_author);  // Pre-initialize local chain
+        
+        Self {
+            store_id,
+            store,
+            chain_manager,
+            node,
+            rx,
+            entry_tx,
+        }
+    }
+
+    /// Run the actor loop - processes commands until Shutdown received
+    /// Uses blocking_recv since redb is sync and we run in spawn_blocking
+    pub fn run(mut self) {
+        while let Some(cmd) = self.rx.blocking_recv() {
+            match cmd {
+                StoreCmd::Get { key, resp } => {
+                    let _ = resp.send(self.store.get(&key));
+                }
+                StoreCmd::GetHeads { key, resp } => {
+                    let _ = resp.send(self.store.get_heads(&key));
+                }
+                StoreCmd::List { include_deleted, resp } => {
+                    let _ = resp.send(self.store.list_all(include_deleted));
+                }
+                StoreCmd::ListByPrefix { prefix, include_deleted, resp } => {
+                    let _ = resp.send(self.store.list_by_prefix(&prefix, include_deleted));
+                }
+                StoreCmd::Put { key, value, resp } => {
+                    let result = self.do_put(&key, &value);
+                    let _ = resp.send(result);
+                }
+                StoreCmd::Delete { key, resp } => {
+                    let result = self.do_delete(&key);
+                    let _ = resp.send(result);
+                }
+                StoreCmd::LogSeq { resp } => {
+                    let local_author = self.node.public_key_bytes();
+                    let len = self.chain_manager.get(&local_author)
+                        .map(|c| c.len())
+                        .unwrap_or(0);
+                    let _ = resp.send(len);
+                }
+                StoreCmd::AppliedSeq { resp } => {
+                    let author = self.node.public_key_bytes();
+                    let result = self.store.author_state(&author)
+                        .map(|s| s.map(|a| a.seq).unwrap_or(0));
+                    let _ = resp.send(result);
+                }
+                StoreCmd::AuthorState { author, resp } => {
+                    let _ = resp.send(self.store.author_state(&author));
+                }
+                StoreCmd::SyncState { resp } => {
+                    let _ = resp.send(self.store.sync_state());
+                }
+                StoreCmd::ReadEntriesAfter { author, from_hash, resp } => {
+                    // Read entries from the log file for this author
+                    let result = self.do_read_entries_after(&author, from_hash);
+                    let _ = resp.send(result);
+                }
+                StoreCmd::ApplyEntry { entry, resp } => {
+                    // Use SigChainManager to append to the correct author's log
+                    if let Err(e) = self.chain_manager.append_entry(&entry) {
+                        let _ = resp.send(Err(StoreError::from(e)));
+                        continue;
+                    }
+                    
+                    // Then apply to store
+                    let result = self.store.apply_entry(&entry);
+                    let _ = resp.send(result);
+                }
+                StoreCmd::LogStats { resp } => {
+                    let _ = resp.send(self.chain_manager.log_stats());
+                }
+                StoreCmd::Shutdown => {
+                    break;
+                }
+            }
+        }
+    }
+
+    fn do_put(&mut self, key: &[u8], value: &[u8]) -> Result<u64, StoreActorError> {
+        let heads = self.store.get_heads(key)?;
+        
+        // Idempotency check (pure function)
+        if !Store::needs_put(&heads, value) {
+            let local_author = self.node.public_key_bytes();
+            return Ok(self.chain_manager.get(&local_author).map(|c| c.len()).unwrap_or(0));
+        }
+        
+        let parent_hashes: Vec<Vec<u8>> = heads.iter().map(|h| h.hash.clone()).collect();
+        self.commit_entry(parent_hashes, |b| b.put(key.to_vec(), value.to_vec()))
+    }
+
+    fn do_delete(&mut self, key: &[u8]) -> Result<u64, StoreActorError> {
+        let heads = self.store.get_heads(key)?;
+        
+        // Idempotency check (pure function)
+        if !Store::needs_delete(&heads) {
+            let local_author = self.node.public_key_bytes();
+            return Ok(self.chain_manager.get(&local_author).map(|c| c.len()).unwrap_or(0));
+        }
+        
+        let parent_hashes: Vec<Vec<u8>> = heads.iter().map(|h| h.hash.clone()).collect();
+        self.commit_entry(parent_hashes, |b| b.delete(key.to_vec()))
+    }
+
+    fn commit_entry<F>(&mut self, parent_hashes: Vec<Vec<u8>>, build: F) -> Result<u64, StoreActorError>
+    where
+        F: FnOnce(EntryBuilder) -> EntryBuilder,
+    {
+        let local_author = self.node.public_key_bytes();
+        let sigchain = self.chain_manager.get_or_create(local_author);
+        
+        let seq = sigchain.len() + 1;
+        let prev_hash = *sigchain.last_hash();
+
+        let builder = EntryBuilder::new(seq, HLC::now())
+            .store_id(self.store_id.as_bytes().to_vec())
+            .prev_hash(prev_hash.to_vec())
+            .parent_hashes(parent_hashes);
+        let entry = build(builder).sign(&self.node);
+
+        // Append to local sigchain
+        let sigchain = self.chain_manager.get_or_create(local_author);
+        sigchain.append(&entry)?;
+        self.store.apply_entry(&entry)?;
+        
+        // Broadcast the entry to listeners (for gossip)
+        let _ = self.entry_tx.send(entry.clone());
+
+        Ok(seq)
+    }
+
+    fn do_read_entries_after(
+        &self,
+        author: &[u8; 32],
+        from_hash: Option<[u8; 32]>,
+    ) -> Result<Vec<SignedEntry>, StoreError> {
+        // Build log path for this author
+        let author_hex = hex::encode(author);
+        let log_path = self.chain_manager.logs_dir().join(format!("{}.log", author_hex));
+        
+        if !log_path.exists() {
+            return Ok(Vec::new());  // No log file for this author
+        }
+        
+        // Use lattice_core's read_entries_after
+        log::read_entries_after(&log_path, from_hash)
+            .map_err(StoreError::from)
+    }
+}
+
+/// Spawn a store actor in a new thread, returns (cmd_tx, entry_tx, join_handle)
+/// Uses std::thread since redb is blocking
+pub fn spawn_store_actor(
+    store_id: Uuid,
+    store: Store,
+    sigchain: SigChain,
+    node: NodeIdentity,
+) -> (mpsc::Sender<StoreCmd>, broadcast::Sender<SignedEntry>, JoinHandle<()>) {
+    let (tx, rx) = mpsc::channel(32);
+    let (entry_tx, _entry_rx) = broadcast::channel(64);
+    let actor = StoreActor::new(store_id, store, sigchain, node, rx, entry_tx.clone());
+    let handle = thread::spawn(move || actor.run());
+    (tx, entry_tx, handle)
+}

lattice-core/src/sync_state.rs

@@ -0,0 +1,325 @@
+//! Sync state for causality tracking and reconciliation
+
+use std::collections::{HashMap, HashSet};
+
+/// Author ID type (32-byte Ed25519 public key)
+pub type Author = [u8; 32];
+
+/// Per-author sync information: seq + all head hashes.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct AuthorInfo {
+    pub seq: u64,
+    pub heads: HashSet<[u8; 32]>,  // All head hashes for this author
+}
+
+impl AuthorInfo {
+    pub fn new(seq: u64, hash: [u8; 32]) -> Self {
+        let mut heads = HashSet::new();
+        heads.insert(hash);
+        Self { seq, heads }
+    }
+    
+    pub fn with_heads(seq: u64, heads: HashSet<[u8; 32]>) -> Self {
+        Self { seq, heads }
+    }
+}
+
+/// Sync state tracking per-author sequence numbers and head hashes.
+///
+/// Used during reconciliation to identify missing entries between peers.
+/// Tracks all head hashes per author to handle forks correctly.
+#[derive(Debug, Clone, Default)]
+pub struct SyncState {
+    authors: HashMap<Author, AuthorInfo>,
+}
+
+/// Describes entries needed from a peer for a specific author.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct MissingRange {
+    pub author: Author,
+    pub from_seq: u64,        // exclusive - we have up to this
+    pub from_hash: [u8; 32],  // hash to resume reading after (zero = start)
+    pub to_seq: u64,          // inclusive - peer has up to this
+}
+
+impl SyncState {
+    /// Create a new empty sync state.
+    pub fn new() -> Self {
+        Self {
+            authors: HashMap::new(),
+        }
+    }
+
+    /// Get the info for an author (returns None if not present).
+    pub fn get(&self, author: &Author) -> Option<&AuthorInfo> {
+        self.authors.get(author)
+    }
+
+    /// Get the sequence number for an author (returns 0 if not present).
+    pub fn seq(&self, author: &Author) -> u64 {
+        self.authors.get(author).map(|i| i.seq).unwrap_or(0)
+    }
+    
+    /// Get head hashes for an author (returns empty set if not present).
+    pub fn heads(&self, author: &Author) -> HashSet<[u8; 32]> {
+        self.authors.get(author).map(|i| i.heads.clone()).unwrap_or_default()
+    }
+
+    /// Set the info for an author (single hash convenience method).
+    pub fn set(&mut self, author: Author, seq: u64, hash: [u8; 32]) {
+        self.authors.insert(author, AuthorInfo::new(seq, hash));
+    }
+    
+    /// Set the info for an author with multiple heads.
+    pub fn set_heads(&mut self, author: Author, seq: u64, heads: HashSet<[u8; 32]>) {
+        self.authors.insert(author, AuthorInfo::with_heads(seq, heads));
+    }
+    
+    /// Add a head hash for an author (updates seq if higher).
+    pub fn add_head(&mut self, author: Author, seq: u64, hash: [u8; 32]) {
+        if let Some(info) = self.authors.get_mut(&author) {
+            info.heads.insert(hash);
+            if seq > info.seq {
+                info.seq = seq;
+            }
+        } else {
+            self.set(author, seq, hash);
+        }
+    }
+
+    /// Get all authors and their info.
+    pub fn authors(&self) -> &HashMap<Author, AuthorInfo> {
+        &self.authors
+    }
+
+    /// Compute what entries we're missing compared to a peer's state.
+    ///
+    /// Returns ranges of entries we need from the peer.
+    /// Compares hash sets when seq matches to detect forks.
+    pub fn diff(&self, peer: &SyncState) -> Vec<MissingRange> {
+        let mut missing = Vec::new();
+
+        for (author, peer_info) in peer.authors() {
+            let my_seq = self.seq(author);
+            let my_heads = self.heads(author);
+            
+            // We need entries if:
+            // 1. Peer's seq is higher than ours, OR
+            // 2. Peer's seq equals ours but they have heads we don't (fork)
+            let need_entries = if peer_info.seq > my_seq {
+                true
+            } else if peer_info.seq == my_seq && my_seq > 0 {
+                // Same seq - check for forks (different hashes at same seq)
+                peer_info.heads.iter().any(|h| !my_heads.contains(h))
+            } else {
+                false
+            };
+            
+            if need_entries {
+                // Request from our common ancestor (or start if we have nothing)
+                let from_hash = if my_heads.is_empty() {
+                    [0u8; 32]
+                } else {
+                    *my_heads.iter().next().unwrap()
+                };
+                
+                missing.push(MissingRange {
+                    author: *author,
+                    from_seq: my_seq,
+                    from_hash,
+                    to_seq: peer_info.seq,
+                });
+            }
+        }
+
+        missing
+    }
+
+    /// Merge another sync state into this one (union of heads, max seq).
+    pub fn merge(&mut self, other: &SyncState) {
+        for (author, info) in other.authors() {
+            if let Some(my_info) = self.authors.get_mut(author) {
+                // Union heads
+                for h in &info.heads {
+                    my_info.heads.insert(*h);
+                }
+                // Take max seq
+                if info.seq > my_info.seq {
+                    my_info.seq = info.seq;
+                }
+            } else {
+                self.authors.insert(*author, info.clone());
+            }
+        }
+    }
+
+    /// Convert to proto message for network transmission
+    pub fn to_proto(&self) -> crate::proto::SyncState {
+        let frontiers = self.authors.iter().map(|(author, info)| {
+            crate::proto::Frontier {
+                author_id: author.to_vec(),
+                max_seq: info.seq,
+                head_hashes: info.heads.iter().map(|h| h.to_vec()).collect(),
+            }
+        }).collect();
+        crate::proto::SyncState {
+            frontiers,
+            sender_hlc: None,
+        }
+    }
+
+    /// Create from proto message
+    pub fn from_proto(proto: &crate::proto::SyncState) -> Self {
+        let mut state = Self::new();
+        for frontier in &proto.frontiers {
+            if frontier.author_id.len() == 32 {
+                let mut author = [0u8; 32];
+                author.copy_from_slice(&frontier.author_id);
+                
+                let mut heads = HashSet::new();
+                for hash_bytes in &frontier.head_hashes {
+                    if hash_bytes.len() == 32 {
+                        let mut hash = [0u8; 32];
+                        hash.copy_from_slice(hash_bytes);
+                        heads.insert(hash);
+                    }
+                }
+                
+                if heads.is_empty() {
+                    // Fallback: empty hash if no heads provided
+                    heads.insert([0u8; 32]);
+                }
+                
+                state.set_heads(author, frontier.max_seq, heads);
+            }
+        }
+        state
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_diff_empty() {
+        let a = SyncState::new();
+        let b = SyncState::new();
+        assert!(a.diff(&b).is_empty());
+    }
+
+    #[test]
+    fn test_diff_peer_ahead() {
+        let mut a = SyncState::new();
+        let mut b = SyncState::new();
+        
+        let author = [1u8; 32];
+        let hash_a = [0xAA; 32];
+        let hash_b = [0xBB; 32];
+        
+        a.set(author, 5, hash_a);
+        b.set(author, 10, hash_b);
+
+        let missing = a.diff(&b);
+        assert_eq!(missing.len(), 1);
+        assert_eq!(missing[0].author, author);
+        assert_eq!(missing[0].from_seq, 5);
+        assert_eq!(missing[0].from_hash, hash_a);  // Resume after our hash
+        assert_eq!(missing[0].to_seq, 10);
+    }
+
+    #[test]
+    fn test_diff_i_am_ahead() {
+        let mut a = SyncState::new();
+        let mut b = SyncState::new();
+        
+        let author = [1u8; 32];
+        a.set(author, 10, [0xAA; 32]);
+        b.set(author, 5, [0xBB; 32]);
+
+        // I'm ahead, so I don't need anything from peer
+        let missing = a.diff(&b);
+        assert!(missing.is_empty());
+    }
+
+    #[test]
+    fn test_diff_new_author() {
+        let a = SyncState::new();
+        let mut b = SyncState::new();
+        
+        let author = [2u8; 32];
+        b.set(author, 3, [0xBB; 32]);
+
+        // Peer has author I don't have
+        let missing = a.diff(&b);
+        assert_eq!(missing.len(), 1);
+        assert_eq!(missing[0].from_seq, 0);
+        assert_eq!(missing[0].from_hash, [0u8; 32]);  // Zero hash = read from start
+        assert_eq!(missing[0].to_seq, 3);
+    }
+
+    #[test]
+    fn test_merge() {
+        let mut a = SyncState::new();
+        let mut b = SyncState::new();
+        
+        let author1 = [1u8; 32];
+        let author2 = [2u8; 32];
+        
+        a.set(author1, 10, [0xA1; 32]);
+        a.set(author2, 5, [0xA2; 32]);
+        
+        b.set(author1, 5, [0xB1; 32]);   // a is ahead
+        b.set(author2, 8, [0xB2; 32]);   // b is ahead
+
+        a.merge(&b);
+        assert_eq!(a.seq(&author1), 10);  // kept a's value
+        assert_eq!(a.seq(&author2), 8);   // took b's value
+    }
+
+    /// This test documents a known issue: SyncState tracks only ONE hash per author,
+    /// but with forks/multi-heads, there could be multiple branches.
+    /// 
+    /// Scenario:
+    /// - Author writes entry1 (hash=A)
+    /// - Two peers independently write entry2 and entry3 (both have prev=A)
+    /// - Peer1 has: entry1 -> entry2 (seq=2, hash=B)
+    /// - Peer2 has: entry1 -> entry3 (seq=2, hash=C)
+    /// - When Peer3 syncs with Peer1, SyncState says "I need entries after hash=B"
+    /// - But Peer2 only has entries after hash=A, so Peer3 never gets entry3!
+    /// 
+    #[test]
+    fn test_multihead_sync_inconsistency() {
+        // This is a conceptual test showing the problem
+        // In reality, both forks would have seq=2 but different hashes
+        // SyncState can only track one, so the other branch gets lost
+        
+        let mut peer1_state = SyncState::new();
+        let mut peer2_state = SyncState::new();
+        let new_peer_state = SyncState::new();
+        
+        let author = [1u8; 32];
+        
+        // Both peers have seq=2, but different hashes (different forks)
+        peer1_state.set(author, 2, [0xBB; 32]); // entry1 -> entry2
+        peer2_state.set(author, 2, [0xCC; 32]); // entry1 -> entry3
+        
+        // New peer syncs with peer1 first
+        let missing_from_peer1 = new_peer_state.diff(&peer1_state);
+        assert_eq!(missing_from_peer1.len(), 1);
+        assert_eq!(missing_from_peer1[0].to_seq, 2);
+        
+        // After applying peer1's entries, new peer has seq=2, hash=BB
+        let mut after_peer1 = new_peer_state.clone();
+        after_peer1.set(author, 2, [0xBB; 32]);
+        
+        // Now sync with peer2 - BUG: new peer thinks it's up to date!
+        let missing_from_peer2 = after_peer1.diff(&peer2_state);
+        
+        // This assertion FAILS - we get empty missing even though peer2 has entry3!
+        // The bug: peer2's seq=2 equals our seq=2, so we think we're in sync
+        // But peer2's hash=0xCC != our hash=0xBB - they have different entries!
+        assert!(!missing_from_peer2.is_empty(), 
+            "BUG: SyncState misses peer2's fork because seq numbers match");
+    }
+}

lattice-core/src/vector_clock.rs

@@ -1,41 +0,0 @@
-//! Vector clocks for causality tracking
-
-use std::collections::HashMap;
-
-/// A vector clock for tracking "how much" of each node's log has been seen.
-///
-/// Used during reconciliation to identify missing entries between peers.
-pub struct VectorClock {
-    clocks: HashMap<[u8; 32], u64>,
-}
-
-impl VectorClock {
-    /// Create a new empty vector clock.
-    pub fn new() -> Self {
-        Self {
-            clocks: HashMap::new(),
-        }
-    }
-
-    /// Get the clock value for a node (returns 0 if not present).
-    pub fn get(&self, node_id: &[u8; 32]) -> u64 {
-        self.clocks.get(node_id).copied().unwrap_or(0)
-    }
-
-    /// Set the clock value for a node.
-    pub fn set(&mut self, node_id: [u8; 32], value: u64) {
-        self.clocks.insert(node_id, value);
-    }
-
-    /// Increment the clock for a node.
-    pub fn increment(&mut self, node_id: [u8; 32]) {
-        let current = self.get(&node_id);
-        self.set(node_id, current + 1);
-    }
-}
-
-impl Default for VectorClock {
-    fn default() -> Self {
-        Self::new()
-    }
-}

lattice-net/Cargo.toml

@@ -9,10 +9,17 @@ license.workspace = true
 lattice-core = { workspace = true }
 iroh = { workspace = true }
 iroh-gossip = { workspace = true }
+prost = { workspace = true }
 tokio = { workspace = true }
 thiserror = { workspace = true }
 tracing = { workspace = true }
 bytes = { workspace = true }
+tokio-util = { workspace = true }
+futures-util = { workspace = true }
+hex = { workspace = true }
+blake3.workspace = true
+anyhow = "1.0.100"
+futures-lite = "2.6.1"
 
 [dev-dependencies]
 tokio-test = { workspace = true }

lattice-net/src/endpoint.rs

@@ -0,0 +1,60 @@
+//! Iroh endpoint for network connectivity
+//!
+//! Creates an Iroh endpoint from the node's Ed25519 secret key,
+//! ensuring the same identity is used for both Lattice and Iroh.
+//! 
+//! Discovery: Uses both DNS (default) and mDNS (local network)
+
+use iroh::{Endpoint, endpoint::{BindError, Connection, ConnectError}};
+use iroh::discovery::mdns::MdnsDiscovery;
+pub use iroh::PublicKey;
+
+/// ALPN protocol identifier for Lattice sync
+pub const LATTICE_ALPN: &[u8] = b"lattice-sync/1";
+
+/// Wrapper around Iroh endpoint with Lattice integration
+pub struct LatticeEndpoint {
+    endpoint: Endpoint,
+}
+
+impl LatticeEndpoint {
+    /// Create a new endpoint from Ed25519 secret key bytes (from identity.key)
+    /// Enables both DNS discovery (internet) and mDNS discovery (local network)
+    pub async fn new(secret_key_bytes: [u8; 32]) -> Result<Self, BindError> {
+        let secret_key = iroh::SecretKey::from_bytes(&secret_key_bytes);
+        
+        // mDNS for local network discovery
+        let mdns = MdnsDiscovery::builder();
+        
+        let endpoint = Endpoint::builder()
+            .secret_key(secret_key)
+            .alpns(vec![
+                LATTICE_ALPN.to_vec(),
+                iroh_gossip::ALPN.to_vec(),  // Also accept gossip protocol
+            ])
+            .discovery(mdns)  // Add mDNS on top of default DNS
+            .bind()
+            .await?;
+        Ok(Self { endpoint })
+    }
+
+    /// Get the public key (same as Lattice pubkey, can be shared with peers)
+    pub fn public_key(&self) -> PublicKey {
+        self.endpoint.secret_key().public()
+    }
+
+    /// Connect to a peer by their public key
+    pub async fn connect(&self, peer: PublicKey) -> Result<Connection, ConnectError> {
+        self.endpoint.connect(peer, LATTICE_ALPN).await
+    }
+
+    /// Accept an incoming connection
+    pub async fn accept(&self) -> Option<iroh::endpoint::Incoming> {
+        self.endpoint.accept().await
+    }
+
+    /// Get the underlying endpoint
+    pub fn endpoint(&self) -> &Endpoint {
+        &self.endpoint
+    }
+}

lattice-net/src/framing.rs

@@ -0,0 +1,63 @@
+//! Message framing for Iroh streams using tokio-util LengthDelimitedCodec
+//!
+//! Provides a clean interface for sending/receiving length-prefixed PeerMessage
+//! over QUIC streams without manual buffer management.
+
+use futures_util::{SinkExt, StreamExt};
+use lattice_core::proto::PeerMessage;
+use prost::Message;
+use tokio_util::codec::{FramedRead, FramedWrite, LengthDelimitedCodec};
+
+/// Framed writer for sending PeerMessage over an Iroh SendStream
+pub struct MessageSink {
+    inner: FramedWrite<iroh::endpoint::SendStream, LengthDelimitedCodec>,
+}
+
+impl MessageSink {
+    pub fn new(stream: iroh::endpoint::SendStream) -> Self {
+        Self {
+            inner: FramedWrite::new(stream, LengthDelimitedCodec::new()),
+        }
+    }
+
+    /// Send a PeerMessage (length-prefixed)
+    pub async fn send(&mut self, msg: &PeerMessage) -> Result<(), String> {
+        let bytes = msg.encode_to_vec();
+        self.inner.send(bytes.into()).await
+            .map_err(|e| format!("Send error: {}", e))
+    }
+
+    /// Finish the stream (signal we're done sending)
+    pub async fn finish(self) -> Result<(), String> {
+        let mut stream = self.inner.into_inner();
+        let _ = stream.finish();
+        stream.stopped().await.ok();
+        Ok(())
+    }
+}
+
+/// Framed reader for receiving PeerMessage from an Iroh RecvStream
+pub struct MessageStream {
+    inner: FramedRead<iroh::endpoint::RecvStream, LengthDelimitedCodec>,
+}
+
+impl MessageStream {
+    pub fn new(stream: iroh::endpoint::RecvStream) -> Self {
+        Self {
+            inner: FramedRead::new(stream, LengthDelimitedCodec::new()),
+        }
+    }
+
+    /// Receive next PeerMessage (or None if stream closed)
+    pub async fn recv(&mut self) -> Result<Option<PeerMessage>, String> {
+        match self.inner.next().await {
+            Some(Ok(bytes)) => {
+                PeerMessage::decode(&bytes[..])
+                    .map(Some)
+                    .map_err(|e| format!("Decode error: {}", e))
+            }
+            Some(Err(e)) => Err(format!("Read error: {}", e)),
+            None => Ok(None),
+        }
+    }
+}

lattice-net/src/lib.rs

@@ -1,8 +1,23 @@
 //! Lattice Networking
 //!
 //! Networking layer using Iroh:
+//! - **Endpoint**: Network identity and connection management
 //! - **Gossip**: Broadcasting changes across the mesh
 //! - **Unicast**: Point-to-point communication for reconciliation
+//! - **Framing**: Length-delimited message framing for QUIC streams
+//! - **Mesh**: Peer-to-peer join and sync operations
 
+pub mod endpoint;
 pub mod gossip;
-pub mod unicast;
+pub mod framing;
+pub mod mesh;
+
+pub use endpoint::{LatticeEndpoint, PublicKey, LATTICE_ALPN};
+pub use framing::{MessageSink, MessageStream};
+pub use lattice_core::proto::{SyncRequest, SyncResponse, SyncEntry, SyncDone, SyncState, Frontier};
+pub use mesh::{LatticeServer, SyncResult};
+
+/// Parse a PublicKey (NodeId) from hex or base32 string
+pub fn parse_node_id(s: &str) -> Result<PublicKey, String> {
+    s.parse().map_err(|e| format!("{}", e))
+}

lattice-net/src/mesh/mod.rs

@@ -0,0 +1,10 @@
+//! Mesh networking - peer-to-peer join and sync operations
+//!
+//! - **server**: LatticeServer for mesh networking (join, sync, accept loop)
+//! - **protocol**: Shared send/receive entry logic
+
+mod server;
+mod protocol;
+
+pub use server::{LatticeServer, SyncResult};
+pub use protocol::{send_missing_entries, receive_entries};

lattice-net/src/mesh/protocol.rs

@@ -0,0 +1,86 @@
+//! Protocol - shared logic for bidirectional sync entry exchange
+
+use crate::{MessageSink, MessageStream};
+use lattice_core::{StoreHandle, CausalEntryIter};
+use lattice_core::proto::{peer_message, PeerMessage, SignedEntry};
+use lattice_core::sync_state::SyncState;
+use prost::Message;
+use std::collections::VecDeque;
+
+/// Send entries that peer is missing based on state diff.
+/// Returns (entries_sent, optional_error).
+pub async fn send_missing_entries(
+    sink: &mut MessageSink,
+    store: &StoreHandle,
+    my_state: &SyncState,
+    peer_state: &SyncState,
+) -> Result<u64, String> {
+    let missing = peer_state.diff(my_state);
+    
+    // Build queues for each author's entries
+    let mut author_entries: Vec<VecDeque<SignedEntry>> = Vec::new();
+    for range in missing {
+        let from_hash = if range.from_hash == [0u8; 32] { None } else { Some(range.from_hash) };
+        let entries = store.read_entries_after(&range.author, from_hash).await
+            .map_err(|e| format!("Failed to read entries: {}", e))?;
+        if !entries.is_empty() {
+            author_entries.push(entries.into());
+        }
+    }
+    
+    // Stream entries in HLC (causal) order
+    let mut entries_sent = 0u64;
+    for entry in CausalEntryIter::new(author_entries) {
+        let sync_msg = PeerMessage {
+            message: Some(peer_message::Message::SyncEntry(lattice_core::proto::SyncEntry {
+                signed_entry: entry.encode_to_vec(),
+                hash: vec![],
+            })),
+        };
+        sink.send(&sync_msg).await?;
+        entries_sent += 1;
+    }
+    
+    // Send SyncDone
+    let done = PeerMessage {
+        message: Some(peer_message::Message::SyncDone(lattice_core::proto::SyncDone {
+            entries_sent,
+        })),
+    };
+    sink.send(&done).await?;
+    
+    Ok(entries_sent)
+}
+
+/// Receive and apply entries until SyncDone is received.
+/// Returns (entries_applied, entries_reported_by_peer).
+pub async fn receive_entries(
+    stream: &mut MessageStream,
+    store: &StoreHandle,
+) -> Result<(u64, u64), String> {
+    let mut entries_applied = 0u64;
+    let mut entries_reported = 0u64;
+    
+    loop {
+        match stream.recv().await {
+            Ok(Some(msg)) => match msg.message {
+                Some(peer_message::Message::SyncEntry(entry)) => {
+                    if let Ok(signed) = SignedEntry::decode(&entry.signed_entry[..]) {
+                        if store.apply_entry(signed).await.is_ok() {
+                            entries_applied += 1;
+                        }
+                    }
+                }
+                Some(peer_message::Message::SyncDone(done)) => {
+                    entries_reported = done.entries_sent;
+                    break;
+                }
+                _ => {}
+            }
+            Ok(None) => break,
+            Err(_) => break,
+        }
+    }
+    
+    Ok((entries_applied, entries_reported))
+}

lattice-net/src/mesh/server.rs

@@ -0,0 +1,575 @@
+//! Server - LatticeServer for mesh networking
+
+use crate::{MessageSink, MessageStream, LatticeEndpoint, parse_node_id, LATTICE_ALPN};
+use lattice_core::{Node, NodeError, NodeEvent, PeerStatus, Uuid, StoreHandle};
+use iroh::endpoint::Connection;
+use iroh::protocol::{Router, ProtocolHandler, AcceptError};
+use iroh_gossip::Gossip;
+use std::sync::Arc;
+use std::collections::HashMap;
+use tokio::sync::RwLock;
+use futures_util::StreamExt;
+use lattice_core::proto::{PeerMessage, peer_message, JoinRequest, JoinResponse, SignedEntry};
+use prost::Message;
+use super::protocol;
+
+/// Result of a sync operation with a peer
+pub struct SyncResult {
+    pub entries_applied: u64,
+    pub entries_sent_by_peer: u64,
+}
+
+/// LatticeServer wraps Node + Endpoint + Gossip and provides mesh networking methods.
+/// Uses Router to handle incoming connections for both sync and gossip protocols.
+pub struct LatticeServer {
+    node: Arc<Node>,
+    endpoint: LatticeEndpoint,
+    gossip: Gossip,
+    #[allow(dead_code)]
+    router: Router,
+    /// Gossip senders per store topic
+    gossip_senders: Arc<RwLock<HashMap<Uuid, iroh_gossip::api::GossipSender>>>,
+}
+
+/// Protocol handler for lattice sync connections
+struct SyncProtocol {
+    node: Arc<Node>,
+}
+
+impl std::fmt::Debug for SyncProtocol {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("SyncProtocol").finish()
+    }
+}
+
+
+impl ProtocolHandler for SyncProtocol {
+    fn accept(&self, conn: Connection) -> impl std::future::Future<Output = Result<(), AcceptError>> + Send {
+        let node = self.node.clone();
+        Box::pin(async move {
+            if let Err(e) = handle_connection(node, conn).await {
+                eprintln!("[Accept] Error: {}", e);
+                // Log error but return Ok - protocol handled the connection
+            }
+            Ok(())
+        })
+    }
+}
+
+impl LatticeServer {
+    /// Create a new LatticeServer from just a Node (creates endpoint internally).
+    pub async fn new_from_node(node: Arc<Node>) -> Result<Self, String> {
+        let endpoint = LatticeEndpoint::new(node.secret_key_bytes()).await
+            .map_err(|e| format!("Failed to create endpoint: {}", e))?;
+        Self::new(node, endpoint).await
+    }
+    
+    /// Create a new LatticeServer with existing endpoint.
+    pub async fn new(node: Arc<Node>, endpoint: LatticeEndpoint) -> Result<Self, String> {
+        // Create gossip instance
+        let gossip = Gossip::builder().spawn(endpoint.endpoint().clone());
+        
+        // Create sync protocol handler
+        let sync_protocol = SyncProtocol { node: node.clone() };
+        
+        // Create router to handle both protocols
+        let router = Router::builder(endpoint.endpoint().clone())
+            .accept(LATTICE_ALPN, sync_protocol)
+            .accept(iroh_gossip::ALPN, gossip.clone())
+            .spawn();
+        
+        let server = Self { 
+            node, 
+            endpoint,
+            gossip,
+            router,
+            gossip_senders: Arc::new(RwLock::new(HashMap::new())),
+        };
+        server.spawn_node_event_listener();
+        
+        // If root store is already open, start gossip for it  
+        if let Some(store) = (*server.node.root_store().await).clone() {
+            println!("[Gossip] Root store already open, starting gossip...");
+            server.join_gossip_topic(store.id()).await?;
+            server.spawn_entry_forward_loop(store);
+        }
+        
+        Ok(server)
+    }
+    
+    /// Spawn a listener for Node events (auto-starts gossip when root store is activated)
+    fn spawn_node_event_listener(&self) {
+        let mut event_rx = self.node.subscribe_events();
+        let gossip_senders = self.gossip_senders.clone();
+        let gossip = self.gossip.clone();
+        let node = self.node.clone();
+        
+        tokio::spawn(async move {
+            while let Ok(event) = event_rx.recv().await {
+                match event {
+                    NodeEvent::RootStoreActivated(store) => {
+                        println!("[Gossip] Root store activated: {}, starting gossip...", store.id());
+                        
+                        let store_id = store.id();
+                        
+                        // Get bootstrap peers from node's peer list
+                        let bootstrap_peers: Vec<iroh::PublicKey> = match node.list_peers().await {
+                            Ok(peers) => {
+                                peers.iter()
+                                    .filter(|p| p.status == PeerStatus::Active)
+                                    .filter_map(|p| parse_node_id(&p.pubkey).ok())
+                                    .collect()
+                            }
+                            Err(e) => {
+                                eprintln!("[Gossip] Failed to list peers: {}, using empty list", e);
+                                Vec::new()
+                            }
+                        };
+                        println!("[Gossip] Bootstrap peers: {}", bootstrap_peers.len());
+                        
+                        // Topic ID from hash of "lattice/{store_id}" for namespacing
+                        let topic_bytes = blake3::hash(format!("lattice/{}", store_id).as_bytes());
+                        let topic_id = iroh_gossip::TopicId::from_bytes(*topic_bytes.as_bytes());
+                        
+                        // Use subscribe (non-blocking) - peers will connect when they sync
+                        // subscribe_and_join would block waiting for peers we can't reach yet
+                        match gossip.subscribe(topic_id, bootstrap_peers).await {
+                            Ok(sub) => {
+                                let (sender, receiver) = sub.split();
+                                
+                                // Store sender
+                                gossip_senders.write().await.insert(store_id, sender);
+                                
+                                // Spawn receive loop
+                                let store_recv = store.clone();
+                                tokio::spawn(async move {
+                                    let mut receiver = receiver;
+                                    println!("[Gossip] Receive loop started for topic {:?}", topic_id);
+                                    
+                                    while let Some(event) = futures_util::StreamExt::next(&mut receiver).await {
+                                        match event {
+                                            Ok(iroh_gossip::api::Event::Received(msg)) => {
+                                                println!("[Gossip] Received {} bytes", msg.content.len());
+                                                if let Ok(entry) = SignedEntry::decode(&msg.content[..]) {
+                                                    if let Err(e) = store_recv.apply_entry(entry).await {
+                                                        eprintln!("[Gossip] Failed to apply entry: {}", e);
+                                                    } else {
+                                                        println!("[Gossip] Applied entry successfully");
+                                                    }
+                                                }
+                                            }
+                                            Ok(other) => {
+                                                println!("[Gossip] Event: {:?}", other);
+                                            }
+                                            Err(e) => {
+                                                eprintln!("[Gossip] Error: {}", e);
+                                            }
+                                        }
+                                    }
+                                });
+                                
+                                // Spawn entry forward loop
+                                let gossip_senders = gossip_senders.clone();
+                                let mut entry_rx = store.subscribe_entries();
+                                tokio::spawn(async move {
+                                    println!("[Gossip] Entry forward loop started for store {}", store_id);
+                                    while let Ok(entry) = entry_rx.recv().await {
+                                        let senders = gossip_senders.read().await;
+                                        if let Some(sender) = senders.get(&store_id) {
+                                            let bytes = entry.encode_to_vec();
+                                            println!("[Gossip] Broadcasting {} bytes", bytes.len());
+                                            let _ = sender.broadcast(bytes.into()).await;
+                                        }
+                                    }
+                                });
+                                
+                                println!("[Gossip] Gossip started for store {}", store_id);
+                            }
+                            Err(e) => {
+                                eprintln!("[Gossip] Failed to subscribe to topic: {}", e);
+                            }
+                        }
+                    }
+                }
+            }
+        });
+    }
+    
+    /// Start gossip for a store (call after store is opened)
+    pub async fn start_gossip_for_store(&self, store: StoreHandle) -> Result<(), String> {
+        println!("[Gossip] Starting gossip for store {}", store.id());
+        self.join_gossip_topic(store.id()).await?;
+        self.spawn_entry_forward_loop(store);
+        Ok(())
+    }
+    
+    /// Spawn a loop that forwards local store entry broadcasts to gossip
+    fn spawn_entry_forward_loop(&self, store: StoreHandle) {
+        let store_id = store.id();
+        let gossip_senders = self.gossip_senders.clone();
+        let mut entry_rx = store.subscribe_entries();
+        
+        println!("[Gossip] Starting entry forward loop for store {}", store_id);
+        
+        tokio::spawn(async move {
+            while let Ok(entry) = entry_rx.recv().await {
+                println!("[Gossip] Received local entry, forwarding to gossip...");
+                // Forward to gossip sender
+                let senders = gossip_senders.read().await;
+                if let Some(sender) = senders.get(&store_id) {
+                    let bytes = entry.encode_to_vec();
+                    println!("[Gossip] Broadcasting {} bytes to topic {}", bytes.len(), store_id);
+                    if let Err(e) = sender.broadcast(bytes.into()).await {
+                        eprintln!("[Gossip] Failed to broadcast entry: {}", e);
+                    } else {
+                        println!("[Gossip] Broadcast successful");
+                    }
+                } else {
+                    eprintln!("[Gossip] No gossip sender for store {}", store_id);
+                }
+            }
+            println!("[Gossip] Entry forward loop ended for store {}", store_id);
+        });
+    }
+    
+    /// Access the underlying node
+    pub fn node(&self) -> &Node {
+        &self.node
+    }
+    
+    /// Access the underlying endpoint
+    pub fn endpoint(&self) -> &LatticeEndpoint {
+        &self.endpoint
+    }
+
+    
+    /// Join gossip topic for a store (subscribes and spawns receive loop)
+    pub async fn join_gossip_topic(&self, store_id: Uuid) -> Result<(), String> {
+        // Get active peers to bootstrap gossip
+        let peers = self.node.list_peers().await
+            .map_err(|e| format!("Failed to list peers: {}", e))?;
+        
+        let bootstrap_peers: Vec<iroh::PublicKey> = peers.iter()
+            .filter(|p| p.status == PeerStatus::Active)
+            .filter_map(|p| parse_node_id(&p.pubkey).ok())
+            .collect();
+        
+        println!("[Gossip] Joining topic {} with {} bootstrap peers", store_id, bootstrap_peers.len());
+        
+        // Topic ID from store UUID bytes (padded to 32 bytes)
+        // Topic ID from hash of "lattice/{store_id}" for namespacing
+        let topic_bytes = blake3::hash(format!("lattice/{}", store_id).as_bytes());
+        let topic_id = iroh_gossip::TopicId::from_bytes(*topic_bytes.as_bytes());
+        
+        // Subscribe to topic
+        let (sender, mut receiver) = self.gossip.subscribe(topic_id, bootstrap_peers).await
+            .map_err(|e| format!("Failed to subscribe to gossip topic: {}", e))?
+            .split();
+        
+        // Store sender for broadcasting
+        {
+            let mut senders = self.gossip_senders.write().await;
+            senders.insert(store_id, sender);
+        }
+        
+        // Spawn receive loop
+        let node = self.node.clone();
+        let topic = topic_id;
+        tokio::spawn(async move {
+            // StreamExt imported at module level
+            println!("[Gossip] Receive loop started for topic {:?}", topic);
+            
+            while let Some(event) = receiver.next().await {
+                match event {
+                    Ok(iroh_gossip::api::Event::Received(message)) => {
+                        println!("[Gossip] Received gossip message: {} bytes", message.content.len());
+                        // Decode SignedEntry and apply
+                        match SignedEntry::decode(&message.content[..]) {
+                            Ok(entry) => {
+                                println!("[Gossip] Decoded entry, applying...");
+                                // Find store and apply entry
+                                if let Some(store) = (*node.root_store().await).clone() {
+                                    if let Err(e) = store.apply_entry(entry.into()).await {
+                                        eprintln!("[Gossip] Failed to apply entry: {}", e);
+                                    } else {
+                                        println!("[Gossip] Entry applied successfully");
+                                    }
+                                } else {
+                                    eprintln!("[Gossip] No root store to apply entry to");
+                                }
+                            }
+                            Err(e) => eprintln!("[Gossip] Failed to decode entry: {}", e),
+                        }
+                    }
+                    Ok(other) => {
+                        println!("[Gossip] Other event: {:?}", other);
+                    }
+                    Err(e) => eprintln!("[Gossip] Receive error: {}", e),
+                }
+            }
+            println!("[Gossip] Receive loop ended for topic");
+        });
+        
+        Ok(())
+    }
+    
+    /// Broadcast an entry to all gossip subscribers for a store
+    pub async fn broadcast_entry(&self, store_id: Uuid, entry: &SignedEntry) -> Result<(), String> {
+        let senders = self.gossip_senders.read().await;
+        if let Some(sender) = senders.get(&store_id) {
+            let bytes = entry.encode_to_vec();
+            sender.broadcast(bytes.into()).await
+                .map_err(|e| format!("Gossip broadcast failed: {}", e))?;
+        }
+        Ok(())
+    }
+    
+    /// Join an existing mesh by connecting to a peer.
+    pub async fn join_mesh(&self, peer_id: iroh::PublicKey) -> Result<StoreHandle, NodeError> {
+        let conn = self.endpoint.connect(peer_id).await
+            .map_err(|e| NodeError::Actor(format!("Connection failed: {}", e)))?;
+        
+        let (send, recv) = conn.open_bi().await
+            .map_err(|e| NodeError::Actor(format!("Failed to open stream: {}", e)))?;
+        
+        let mut sink = MessageSink::new(send);
+        let mut stream = MessageStream::new(recv);
+        
+        // Send JoinRequest
+        let req = PeerMessage {
+            message: Some(peer_message::Message::JoinRequest(JoinRequest {
+                node_pubkey: self.node.node_id().to_vec(),
+            })),
+        };
+        sink.send(&req).await.map_err(|e| NodeError::Actor(e))?;
+        sink.finish().await.map_err(|e| NodeError::Actor(e))?;
+        
+        // Receive JoinResponse
+        let msg = stream.recv().await
+            .map_err(|e| NodeError::Actor(e))?
+            .ok_or_else(|| NodeError::Actor("Peer closed stream".to_string()))?;
+        
+        match msg.message {
+            Some(peer_message::Message::JoinResponse(resp)) => {
+                let store_uuid = lattice_core::Uuid::from_slice(&resp.store_uuid)
+                    .map_err(|_| NodeError::Actor("Invalid UUID from peer".to_string()))?;
+                
+                let handle = self.node.complete_join(store_uuid).await?;
+                
+                // Sync with peer to get initial data
+                println!("[Join] Syncing with peer to get initial data...");
+                if let Ok(result) = self.sync_with_peer(&handle, peer_id).await {
+                    println!("[Join] Initial sync complete: {} entries", result.entries_applied);
+                }
+                
+                Ok(handle)
+            }
+            _ => Err(NodeError::Actor("Unexpected response".to_string())),
+        }
+    }
+    
+    /// Sync with a specific peer.
+    pub async fn sync_with_peer(&self, store: &StoreHandle, peer_id: iroh::PublicKey) -> Result<SyncResult, NodeError> {
+        let conn = self.endpoint.connect(peer_id).await
+            .map_err(|e| NodeError::Actor(format!("Connection failed: {}", e)))?;
+        
+        let (send, recv) = conn.open_bi().await
+            .map_err(|e| NodeError::Actor(format!("Failed to open stream: {}", e)))?;
+        
+        let mut sink = MessageSink::new(send);
+        let mut stream = MessageStream::new(recv);
+        
+        let my_state = store.sync_state().await?;
+        
+        // Send SyncRequest
+        let req = PeerMessage {
+            message: Some(peer_message::Message::SyncRequest(lattice_core::proto::SyncRequest {
+                store_id: store.id().as_bytes().to_vec(),
+                state: Some(my_state.to_proto()),
+                full_sync: false,
+            })),
+        };
+        sink.send(&req).await.map_err(|e| NodeError::Actor(e))?;
+        
+        // Receive SyncResponse
+        let resp_msg = stream.recv().await.map_err(|e| NodeError::Actor(e))?
+            .ok_or_else(|| NodeError::Actor("Peer closed stream".to_string()))?;
+        
+        let peer_state = match resp_msg.message {
+            Some(peer_message::Message::SyncResponse(resp)) => {
+                resp.state.map(|s| lattice_core::sync_state::SyncState::from_proto(&s))
+                    .unwrap_or_default()
+            }
+            _ => return Err(NodeError::Actor("Expected SyncResponse".to_string())),
+        };
+        
+        // Exchange entries
+        let _entries_sent = protocol::send_missing_entries(&mut sink, store, &my_state, &peer_state).await
+            .map_err(|e| NodeError::Actor(e))?;
+        
+        let (entries_applied, entries_sent_by_peer) = protocol::receive_entries(&mut stream, store).await
+            .map_err(|e| NodeError::Actor(e))?;
+        
+        sink.finish().await.map_err(|e| NodeError::Actor(e))?;
+        
+        Ok(SyncResult { entries_applied, entries_sent_by_peer })
+    }
+    
+    /// Sync with all active peers.
+    pub async fn sync_all(&self, store: &StoreHandle) -> Result<Vec<SyncResult>, NodeError> {
+        let peers = self.node.list_peers().await?;
+        let mut results = Vec::new();
+        let my_pubkey = self.endpoint.public_key();
+        
+        for peer in peers {
+            if peer.status != PeerStatus::Active {
+                continue;
+            }
+            
+            let peer_id = match parse_node_id(&peer.pubkey) {
+                Ok(id) => id,
+                Err(e) => {
+                    eprintln!("[Sync] Failed to parse peer {}: {}", peer.pubkey, e);
+                    continue;
+                }
+            };
+            
+            // Skip self
+            if peer_id == my_pubkey {
+                continue;
+            }
+            
+            println!("[Sync] Syncing with {}...", peer_id.fmt_short());
+            match self.sync_with_peer(store, peer_id).await {
+                Ok(result) => {
+                    println!("[Sync] Applied {} entries", result.entries_applied);
+                    results.push(result);
+                }
+                Err(e) => eprintln!("[Sync] Failed: {}", e),
+            }
+        }
+        
+        Ok(results)
+    }
+}
+
+// --- Connection handling ---
+// --- Connection handling ---
+
+/// Handle a single incoming connection
+async fn handle_connection(
+    node: Arc<Node>,
+    conn: Connection,
+) -> Result<(), String> {
+    let remote_id = conn.remote_id();
+    let remote_hex = hex::encode(remote_id.as_bytes());
+    println!("\n[Incoming] {} (ALPN: {})", remote_id.fmt_short(), String::from_utf8_lossy(conn.alpn()));
+    
+    // Parse remote pubkey
+    let remote_pubkey: [u8; 32] = hex::decode(&remote_hex)
+        .map_err(|_| "Invalid pubkey hex")?
+        .try_into()
+        .map_err(|_| "Invalid pubkey length")?;
+    
+    let (send, recv) = conn.accept_bi().await
+        .map_err(|e| format!("Accept stream error: {}", e))?;
+    
+    let sink = MessageSink::new(send);
+    let mut stream = MessageStream::new(recv);
+    
+    // Read first message to determine request type
+    let msg = stream.recv().await?
+        .ok_or_else(|| "Peer closed stream".to_string())?;
+    
+    match msg.message {
+        Some(peer_message::Message::JoinRequest(req)) => {
+            handle_join_request(&node, &remote_pubkey, req, sink).await
+        }
+        Some(peer_message::Message::SyncRequest(req)) => {
+            handle_sync_request(&node, &remote_pubkey, req, sink, stream).await
+        }
+        _ => Err("Unexpected message type".to_string()),
+    }
+}
+
+/// Handle a join request from an invited peer
+async fn handle_join_request(
+    node: &Node,
+    remote_pubkey: &[u8; 32],
+    req: lattice_core::proto::JoinRequest,
+    mut sink: MessageSink,
+) -> Result<(), String> {
+    println!("[Join] Got JoinRequest from {}", hex::encode(&req.node_pubkey));
+    
+    // Accept the join - verifies invited, sets active, returns store ID
+    let acceptance = node.accept_join(remote_pubkey).await
+        .map_err(|e| e.to_string())?;
+    
+    let resp = PeerMessage {
+        message: Some(peer_message::Message::JoinResponse(JoinResponse {
+            store_uuid: acceptance.store_id.as_bytes().to_vec(),
+            inviter_pubkey: vec![],
+        })),
+    };
+    sink.send(&resp).await?;
+    sink.finish().await?;
+    
+    println!("[Join] Sent JoinResponse, peer now active");
+    Ok(())
+}
+
+/// Handle a sync request - bidirectional exchange of entries
+async fn handle_sync_request(
+    node: &Node,
+    remote_pubkey: &[u8; 32],
+    peer_request: lattice_core::proto::SyncRequest,
+    mut sink: MessageSink,
+    mut stream: MessageStream,
+) -> Result<(), String> {
+    // Verify peer is active (allowed to sync)
+    node.verify_peer_status(remote_pubkey, &[PeerStatus::Active]).await
+        .map_err(|e| e.to_string())?;
+    println!("[Sync] Verified peer as active");
+    
+    // Parse store_id from request
+    let store_id = Uuid::from_slice(&peer_request.store_id)
+        .map_err(|_| format!("Invalid store_id in SyncRequest: {} bytes, expected 16", peer_request.store_id.len()))?;
+    
+    println!("[Sync] Received SyncRequest for store {}", store_id);
+    
+    // Open the requested store (uses cache if available)
+    let (store, _info) = node.open_store(store_id).await
+        .map_err(|e| format!("Failed to open store {}: {}", store_id, e))?;
+    
+    println!("[Sync] Received SyncRequest");
+    
+    // Get our sync state
+    let my_state = store.sync_state().await
+        .map_err(|e| format!("Failed to get sync state: {}", e))?;
+    
+    // 1. Send our sync state as response
+    let resp = PeerMessage {
+        message: Some(peer_message::Message::SyncResponse(lattice_core::proto::SyncResponse {
+            store_id: store.id().as_bytes().to_vec(),
+            state: Some(my_state.to_proto()),
+        })),
+    };
+    sink.send(&resp).await?;
+    
+    // 2. Send entries peer is missing
+    let peer_state = peer_request.state
+        .map(|s| lattice_core::sync_state::SyncState::from_proto(&s))
+        .unwrap_or_default();
+    
+    let entries_sent = protocol::send_missing_entries(&mut sink, &store, &my_state, &peer_state).await?;
+    println!("[Sync] Sent {} entries, now receiving from peer...", entries_sent);
+    
+    // 3. Receive entries from requester (bidirectional)
+    let (entries_applied, _) = protocol::receive_entries(&mut stream, &store).await?;
+    
+    sink.finish().await?;
+    
+    println!("[Sync] Applied {} entries from peer", entries_applied);
+    Ok(())
+}
+

lattice-net/src/unicast.rs

@@ -1,3 +0,0 @@
-//! Unicast communication for direct peer-to-peer messaging
-
-// TODO: Implement unicast using iroh

lattice-store/Cargo.toml

@@ -1,15 +0,0 @@
-[package]
-name = "lattice-store"
-description = "Log-based KV store with snapshots and watermarks"
-version.workspace = true
-edition.workspace = true
-license.workspace = true
-
-[dependencies]
-lattice-core = { workspace = true }
-tokio = { workspace = true }
-thiserror = { workspace = true }
-tracing = { workspace = true }
-
-[dev-dependencies]
-tokio-test = { workspace = true }

lattice-store/src/lib.rs

@@ -1,12 +0,0 @@
-//! Lattice Store
-//!
-//! Log-based Key-Value store:
-//! - State is a "view" generated by replaying entries
-//! - Watermarks for agreeing on safe compaction points
-//! - Snapshots for replacing old logs
-
-pub mod store;
-pub mod snapshot;
-pub mod watermark;
-
-pub use store::Store;

lattice-store/src/snapshot.rs

@@ -1,6 +0,0 @@
-//! Snapshots for log compaction
-
-/// A verified snapshot that can replace old log entries.
-pub struct Snapshot {
-    // TODO: snapshot data, watermark, verification
-}

lattice-store/src/store.rs

@@ -1,35 +0,0 @@
-//! The main KV store
-
-use std::collections::HashMap;
-
-/// A log-based Key-Value store.
-///
-/// The store is a dynamic "view" generated by replaying entries.
-pub struct Store {
-    data: HashMap<Vec<u8>, Vec<u8>>,
-}
-
-impl Store {
-    /// Create a new empty store.
-    pub fn new() -> Self {
-        Self {
-            data: HashMap::new(),
-        }
-    }
-
-    /// Get a value by key.
-    pub fn get(&self, key: &[u8]) -> Option<&[u8]> {
-        self.data.get(key).map(|v| v.as_slice())
-    }
-
-    /// Set a value (this would normally go through the log).
-    pub fn set(&mut self, key: Vec<u8>, value: Vec<u8>) {
-        self.data.insert(key, value);
-    }
-}
-
-impl Default for Store {
-    fn default() -> Self {
-        Self::new()
-    }
-}

lattice-store/src/watermark.rs

@@ -1,9 +0,0 @@
-//! Watermarks for agreeing on safe compaction points
-
-/// A watermark representing a safe cut-off point for log compaction.
-///
-/// Nodes use watermarks to agree on which entries can be safely
-/// deleted and replaced with snapshots.
-pub struct Watermark {
-    // TODO: watermark data
-}

proto/lattice.proto

@@ -20,15 +20,42 @@ message Entry {
   // Versioning allows us to change the format radically later if needed
   uint32 version = 1;
 
+  // Store this entry belongs to (16-byte UUID)
+  bytes store_id = 6;
+
   // Ordering Metadata
-  bytes prev_hash = 2;   // Link to previous entry (32 bytes)
+  bytes prev_hash = 2;   // Link to previous sigchain entry (32 bytes)
   uint64 seq = 3;        // Monotonic sequence number
   HLC timestamp = 4;     // Hybrid Logical Clock
 
+  // DAG ancestry: hashes of entries this supersedes (separate from sigchain)
+  repeated bytes parent_hashes = 7;
+
   // The Batch of Operations
   repeated Operation ops = 5;
 }
 
+// HeadInfo: a tip/head in the DAG for a key
+message HeadInfo {
+  bytes value = 1;       // The value at this head
+  uint64 hlc = 2;        // Combined HLC for ordering (wall_time_ms << 16 | counter)
+  bytes author = 3;      // Author's public key (32 bytes)
+  bytes hash = 4;        // Hash of the SignedEntry that created this head
+  bool tombstone = 5;    // True if this head represents a delete
+}
+
+// HeadList: wrapper for storing multiple heads per key in state.db
+message HeadList {
+  repeated HeadInfo heads = 1;
+}
+
+// AuthorState: tracks last applied entry per author for replay optimization
+message AuthorState {
+  uint64 seq = 1;        // Last applied seq for this author's sigchain
+  bytes hash = 2;        // Hash of last applied entry
+  uint64 log_offset = 3; // Byte offset in log file for fast resume
+}
+
 // Hybrid Logical Clock
 message HLC {
   uint64 wall_time = 1;  // Unix timestamp (ms)
@@ -46,12 +73,12 @@ message Operation {
 }
 
 message PutOp {
-  string key = 1;
+  bytes key = 1;
-  bytes value = 2; // Raw bytes allows storing images, JSON, binary, etc.
+  bytes value = 2;
 }
 
 message DeleteOp {
-  string key = 1;
+  bytes key = 1;
 }
 
 // 4. The Sync Handshake (Vector Clocks)
@@ -63,7 +90,7 @@ message SyncState {
 message Frontier {
   bytes author_id = 1;  // Ed25519 public key (32 bytes)
   uint64 max_seq = 2;   // Highest sequence number seen from this author
-  bytes last_hash = 3;  // Hash of the last entry (for chain verification)
+  repeated bytes head_hashes = 3;  // All head hashes for this author
 }
 
 // 5. Log File Record (wrapper for storage)
@@ -71,3 +98,46 @@ message LogRecord {
   bytes hash = 1;         // BLAKE3 hash of entry_bytes (32 bytes)
   bytes entry_bytes = 2;  // Serialized SignedEntry
 }
+
+// 6. Join Protocol Messages (new node joining existing mesh)
+message JoinRequest {
+  bytes node_pubkey = 1;  // Joining node's public key (32 bytes)
+}
+
+message JoinResponse {
+  bytes store_uuid = 1;   // Root store UUID (16 bytes) for new node to create
+  bytes inviter_pubkey = 2;  // Inviter's public key for verification
+}
+
+// 7. Sync Protocol Messages (bidirectional sync after join)
+message SyncRequest {
+  bytes store_id = 1;     // Store UUID to sync (16 bytes)
+  SyncState state = 2;    // Sender's sync state (for incremental sync)
+  bool full_sync = 3;     // If true, request all entries (for join)
+}
+
+message SyncResponse {
+  bytes store_id = 1;     // Store UUID being synced (16 bytes)
+  SyncState state = 2;    // Responder's sync state
+}
+
+message SyncEntry {
+  bytes signed_entry = 1; // Serialized SignedEntry
+  bytes hash = 2;         // Hash for verification
+}
+
+message SyncDone {
+  uint64 entries_sent = 1;
+}
+
+// 8. Peer Message Wrapper (for proper message type discrimination)
+message PeerMessage {
+  oneof message {
+    JoinRequest join_request = 1;
+    JoinResponse join_response = 2;
+    SyncRequest sync_request = 3;
+    SyncResponse sync_response = 4;
+    SyncEntry sync_entry = 5;
+    SyncDone sync_done = 6;
+  }
+}