@agentmemory/agentmemory 0.9.30 → 0.9.31

package/.env.example

@@ -5,9 +5,9 @@
 # Copy this file to `~/.agentmemory/.env` (or to your project root if you
 # prefer scoped config) and uncomment the lines you want to override.
 #
-# Every line is OFF by default — `agentmemory` runs out of the box with no
-# LLM key, no embedding key, and no API auth. Set keys here only when you
-# want to enable the corresponding feature.
+# Optional providers and features are off by default — `agentmemory` runs
+# without an LLM or embedding key. Local API authentication uses a generated
+# secret unless you set AGENTMEMORY_SECRET explicitly.
 #
 # Run `npx -y @agentmemory/agentmemory@latest init` to copy this file into
 # place automatically. Run `npx -y @agentmemory/agentmemory@latest doctor`
@@ -108,10 +108,10 @@
 # -----------------------------------------------------------------------------
 #
 # Bearer-token auth for the REST API + viewer + all integration plugins.
-# Without a secret, REST endpoints are open on loopback. Set this when
-# you expose the daemon beyond loopback or run behind a reverse proxy.
+# The CLI generates a secret when unset and local clients retrieve it from
+# the data directory. Set this explicitly for remote clients/deployments.
 
-# AGENTMEMORY_SECRET=your-secret   # generated into ~/.agentmemory/secret on first start when unset
+# AGENTMEMORY_SECRET=your-secret   # generated into the selected data directory on first start when unset
 # VIEWER_ALLOWED_ORIGINS=https://memory.example.com
 # AGENTMEMORY_IMPORT_ROOT=~/projects
 
@@ -122,7 +122,7 @@
 # BM25_WEIGHT=0.4                                # Hybrid search weight for BM25 leg
 # VECTOR_WEIGHT=0.6                              # Hybrid search weight for vector leg
 # AGENTMEMORY_GRAPH_WEIGHT=0.2                   # Graph traversal bonus on smart-search ranking
-# AGENTMEMORY_GRAPH_COMPACT_ON_BOOT=false       # Skip the one-time background trim of oversized graph provenance at startup
+# AGENTMEMORY_GRAPH_COMPACT_ON_BOOT=false       # Report boot compaction as off (otherwise deferred; manual maintenance only)
 # TOKEN_BUDGET=2000                              # Max tokens injected via mem::context per session
 # MAX_OBS_PER_SESSION=500                        # Per-session observation cap before consolidation kicks in
 # SUMMARIZE_CHUNK_SIZE=400                       # When mem::summarize sees a session larger than this, it chunks observations and map-reduces (chunk-summarize → reduce-merge) to stay within the LLM's context window. Default 400 ≈ 50k tokens per chunk at ~110 tok/obs. Native sessions are capped by MAX_OBS_PER_SESSION; chunking primarily matters for bulk-imported jsonl sessions, which bypass that cap.
@@ -133,11 +133,11 @@
 # -----------------------------------------------------------------------------
 
 # AGENTMEMORY_AUTO_COMPRESS=true                 # Run LLM-written compression on every observation batch. Requires BOTH this flag and a configured provider; a provider alone leaves synthetic compression active.
-# AGENTMEMORY_INJECT_CONTEXT=true                # Inject recalled memories back into agent prompts. Default off — hooks capture observations but do not modify conversation.
-# CONSOLIDATION_ENABLED=true                     # Run the 4-tier consolidation pipeline (memories → semantic → procedural). Default off — opt in once you've measured the LLM cost.
+# AGENTMEMORY_INJECT_CONTEXT=true                # Opt in to Claude-style hook context injection. Pi/OMP use 1 in their launch environment for file enrichment; native prompt recall is independent.
+# CONSOLIDATION_ENABLED=false                    # Disable the consolidation pipeline. Otherwise enabled when an LLM provider is configured, unless AGENTMEMORY_PROVIDER=noop.
 # CONSOLIDATION_DECAY_DAYS=30                    # Age (days) after which non-reinforced memories decay during consolidation
 # GRAPH_EXTRACTION_ENABLED=true                  # Extract concept-graph edges on remember; powers the graph-traversal recall path
-# GRAPH_EXTRACTION_BATCH_SIZE=8                  # Memories per graph-extraction batch
+# GRAPH_EXTRACTION_BATCH_SIZE=10                 # Memories per graph-extraction batch
 
 # Local reasoning models only: set to 1 to ask the model to skip its
 # hidden thinking pass during graph extraction. Extraction runs several
@@ -145,7 +145,7 @@
 # the model think (default).
 # AGENTMEMORY_LLM_NOTHINK=1
 
-# AGENTMEMORY_REFLECT=true                       # Periodically auto-synthesize lessons from memories
+# AGENTMEMORY_REFLECT=true                       # Enable slot reflection; requires AGENTMEMORY_SLOTS=true
 # AGENTMEMORY_DROP_STALE_INDEX=true              # Drop on-disk BM25 / vector index on startup if dim guard fires. Recovery toggle for stuck-state debugging.
 # AGENTMEMORY_IMAGE_EMBEDDINGS=true              # Enable image embeddings when an image provider is present (experimental).
 # AGENTMEMORY_AUDIT_RETENTION_MONTHS=6           # Drop whole audit-log month scopes older than N months on a daily sweep. Default 0 = keep audit history forever.
@@ -155,37 +155,39 @@
 # AGENTMEMORY_VECTOR_BUCKET_SIZE=500              # Vectors held per storage bucket before a new bucket opens; a save only rewrites buckets it changed.
 # AGENTMEMORY_VECTOR_BACKFILL_MAX=500             # Batch size for re-embedding vectors missing since the last save (continues across restarts); also caps the whole-store opt-in per boot.
 # AGENTMEMORY_VECTOR_BACKFILL=all                 # Required to backfill an entire store's vectors at once (no prior vector save, or a stale legacy manifest); unset means that case waits for an explicit opt-in.
-# AGENTMEMORY_VIEWER_STREAM_MAX=500              # Max items kept in the persisted "viewer" live-stream group; older entries are pruned in batches. Floored at 200 (the viewer's reload window); negative values fall back to the default
 
 # -----------------------------------------------------------------------------
 # 6. CLI / runtime knobs
 # -----------------------------------------------------------------------------
 
 # AGENTMEMORY_TOOLS=core                         # all (54 tools, default) | core (8 tools): surface exposed to MCP clients
-# AGENTMEMORY_SLOTS=memory                       # Comma-separated plugin slot names the CLI should claim
+# AGENTMEMORY_SLOTS=true                         # Enable structured memory slots (default off)
 # AGENTMEMORY_DEBUG=1                            # Trace MCP shim probe + standalone fallback decisions to stderr
 # AGENTMEMORY_FORCE_PROXY=1                      # Skip the MCP shim livez probe and trust AGENTMEMORY_URL (for sandboxed MCP clients that can't reach localhost)
 # AGENTMEMORY_PROBE_TIMEOUT_MS=2000              # MCP shim livez probe timeout
+# AGENTMEMORY_READY_TIMEOUT_MS=300000            # CLI worker readiness deadline, including state loading (5 minutes)
+# AGENTMEMORY_ENGINE_READY_TIMEOUT_MS=120000     # CLI engine readiness deadline (2 minutes)
+# Startup deadlines accept whole milliseconds from 1 to 3600000. Shell
+# environment overrides this file. Engine and worker waits have separate
+# budgets; increase them only for a slow store.
+# AGENTMEMORY_ENGINE_MANAGES_WORKER=true         # Custom engine-managed worker wrapper; unset detects direct worker commands
 # AGENTMEMORY_URL=http://localhost:3111          # REST base URL — honored by status, doctor, MCP shim
 # AGENTMEMORY_VIEWER_URL=http://localhost:3113   # Override the viewer URL printed by `agentmemory status`
-# Persistent iii state; same override as `--data-dir`, used by native and
-# Docker starts. Explicit flag/env wins. Otherwise instance 0 reuses an
-# existing ./data/state_store.db or ./data/iii-config.yaml before defaults: macOS
+# Persistent iii state; same override as `--data-dir`. Instance 0 reuses
+# an existing ./data/state_store.db before platform defaults: macOS
 # ~/Library/Application Support/agentmemory; Linux $XDG_DATA_HOME/agentmemory
 # or ~/.local/share/agentmemory; Windows %APPDATA%\agentmemory.
 # AGENTMEMORY_DATA_DIR=/absolute/path
-# AGENTMEMORY_III_CONFIG=/path/to/iii-config.yaml  # Engine config the CLI starts from; the bundled default binds 127.0.0.1
-# AGENTMEMORY_STATE_SAVE_INTERVAL_MS=2000    # How often the engine writes state to disk; a crash loses at most this much
-# AGENTMEMORY_USE_DOCKER=1                      # Prefer bundled Docker even when a compatible native iii binary exists
-# AGENTMEMORY_EXPORT_ROOT=~/agentmemory-backup   # Default destination for `agentmemory export`
+# AGENTMEMORY_ENV_FILE=/absolute/path/to/.env      # Config file override (default ~/.agentmemory/.env)
+# AGENTMEMORY_EXPORT_ROOT=~/agentmemory-backup   # Root directory for Obsidian vault exports
 
-# AGENTMEMORY_STATE_BACKEND=redis                # file (default) | redis — storage adapter for iii-state and iii-stream, see README "Storage backend"
+# AGENTMEMORY_STATE_BACKEND=redis                # file (default) | redis — see docs/reliability/state-storage.md
 # AGENTMEMORY_REDIS_URL=redis://localhost:6379   # Required when AGENTMEMORY_STATE_BACKEND=redis
 
 # STANDALONE_MCP=1                               # MCP shim only — bypass the worker and run @agentmemory/mcp in-process
 # STANDALONE_PERSIST_PATH=~/.agentmemory/local.db  # Path used by the standalone MCP shim's local fallback store
 
-# Snapshot exporter — periodic snapshots of state_store + stream_store.
+# Snapshot exporter — periodic snapshots of state_store.
 # SNAPSHOT_ENABLED=true
 # SNAPSHOT_DIR=~/.agentmemory/snapshots
 # SNAPSHOT_INTERVAL=3600                         # Seconds between snapshots
@@ -200,24 +202,20 @@
 # -----------------------------------------------------------------------------
 
 # III_REST_PORT=3111                             # REST API + MCP HTTP anchor
-# III_STREAM_PORT=3112                           # Preferred streams override; defaults to REST + 1
-# III_STREAMS_PORT=3112                          # Legacy alias for III_STREAM_PORT
 # III_VIEWER_PORT=3113                           # Viewer bind port; defaults to REST + 2
 # III_ENGINE_PORT=49134                          # Engine bind port; defaults to REST + 46023
 # III_ENGINE_URL=ws://localhost:49134            # Explicit/remote worker WebSocket URL; takes precedence over derived engine port
 # Derivation applies only while the corresponding explicit port/URL is unset.
-# `--instance 1` selects 3211/3212/3213/49234 and an instance-1 data subdir.
+# `--instance 1` selects REST/viewer/engine ports 3211/3213/49234 and an instance-1 data subdir.
 
 # -----------------------------------------------------------------------------
 # 8. iii engine pin
 # -----------------------------------------------------------------------------
 #
-# agentmemory pins iii-engine to the same release as its iii-sdk dependency
-# (v0.22.1). The worker speaks the wire protocol of that engine, so the two move
-# together in agentmemory releases. Override with AGENTMEMORY_III_VERSION only
-# if you run your own engine and know it matches.
-
-# AGENTMEMORY_III_VERSION=0.22.1
+# agentmemory pins iii-engine and its iii-sdk dependencies to v0.25.0.
+# The CLI downloads the engine into ~/.agentmemory/bin on macOS/Linux.
+# Native Windows requires installing the matching iii.exe manually.
+# AGENTMEMORY_III_VERSION=0.25.0                # Override only when deliberately matching a different engine
 
 # Engine usage telemetry (iii). Off unless you set it; true opts in.
 # III_TELEMETRY_ENABLED=false
@@ -229,14 +227,22 @@
 # Hourly backstop that marks stale active sessions as abandoned.
 # AGENTMEMORY_SESSION_SWEEP_ENABLED=false        # Disable the hourly sweep
 # AGENTMEMORY_SESSION_SWEEP_STALE_HOURS=24       # Inactivity hours before a session is abandoned
+# AGENTMEMORY_FINALIZE_IDLE_MS=120000          # Idle time after a per-turn Stop before the session is completed and summarized (0 = on every Stop)
 
 # -----------------------------------------------------------------------------
 # 10. Claude Code bridge (opt-in)
 # -----------------------------------------------------------------------------
 
-# CLAUDE_MEMORY_BRIDGE=true                      # Mirror compressed memories into Claude Code's CLAUDE.md
+# CLAUDE_MEMORY_BRIDGE=true                      # Export project memories into Claude Code's MEMORY.md
 # CLAUDE_PROJECT_PATH=/path/to/your/project      # Required when CLAUDE_MEMORY_BRIDGE=true
-# CLAUDE_MEMORY_LINE_BUDGET=200                  # Lines of memory CLAUDE.md should hold
+# CLAUDE_MEMORY_LINE_BUDGET=200                  # Maximum lines in the generated MEMORY.md
+# The bridge creates MEMORY.md only when absent. Every existing file is
+# preserved, including old generated exports and files edited by Claude Code.
+# An identical export is a no-op; different content returns an error. To
+# regenerate, deliberately move the existing file to a backup location first
+# (or remove it if no longer needed), then run memory_claude_bridge_sync again.
+# Only memories scoped to this project are exported; unscoped memories are
+# excluded. Reading MEMORY.md remains available regardless of who wrote it.
 
 # -----------------------------------------------------------------------------
 # 11. Obsidian export (opt-in)

package/AGENTS.md

@@ -4,8 +4,8 @@
 
 agentmemory is a persistent memory system for AI coding agents, built on iii-engine's three primitives (Worker/Function/Trigger). Everything goes through `registerFunction`/`registerTrigger`/`sdk.trigger()` — never bypass iii-engine with standalone SQLite or in-process alternatives.
 
-- **Engine**: iii-sdk 0.22.1 with @iii-dev/helpers 0.22.1 (WebSocket to iii-engine 0.22.1 on port 49134; the client type is `IIIClient`, HTTP requests are `HttpRequest` from `@iii-dev/helpers/http`)
-- **State**: File-based SQLite via iii-engine's StateModule (`./data/state_store.db`)
+- **Engine**: iii-sdk 0.25.0 with @iii-dev/helpers 0.25.0 (Worker Compose-managed iii-engine 0.25.0 on port 49134; the client type is `IIIClient`, HTTP requests are `HttpRequest` from `@iii-dev/helpers/http`)
+- **State**: iii state worker with a file-backed KV store by default (`./data/state_store.db`); Docker pins `save_interval_ms` to 2000 ms
 - **Build**: TypeScript → ESM via tsdown, output to `dist/`
 - **Test**: vitest (`npm test` excludes integration tests)
 
@@ -101,17 +101,17 @@ Hook scripts in `src/hooks/` are standalone Node.js scripts (no iii-sdk import).
 
 ## Testing
 
-- Before every commit run, in order: `npm run build`, `npm run skills:gen`, `npm run docs:sync`, `npm run skills:check`, `npm test` (2,500+ tests). Commit anything the generators changed. CI runs the same build, skills check and tests.
+- Before every commit run, in order: `npm run build`, `npm run skills:gen`, `npm run docs:sync`, `npm run skills:check`, `npm test` (4,000+ tests). Commit anything the generators changed. CI runs the same build, skills check and tests.
 - Before any npm publish: `npm run release:gate` must pass. It installs the packed tarballs into a clean home and runs capture, recovery, restart, export/import, MCP and status checks against the installed CLI (see CONTRIBUTING.md, Release process)
 - Mock pattern: `vi.mock("iii-sdk")` with mock `sdk.trigger`, `kv.get/set/list`
 - Test files go in `test/` with `.test.ts` extension
 - Follow existing patterns in `test/crystallize.test.ts` for function tests
 
-## Current Stats (v0.9.30)
+## Current Stats (v0.9.31)
 
-- 54 MCP tools (8 visible by default, `AGENTMEMORY_TOOLS=all` for all)
-- 138 REST endpoints
+- 54 MCP tools visible by default; `AGENTMEMORY_TOOLS=core` selects 8 essentials
+- 146 REST endpoints
 - 6 MCP resources, 3 MCP prompts
 - 12 hooks, 17 skills
-- 311+ iii functions
-- 2,500+ tests
+- 322+ iii functions
+- 4,000+ tests