Skip to content
This repository was archived by the owner on Nov 16, 2026. It is now read-only.
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
25 commits
Select commit Hold shift + click to select a range
2973f9f
fix: preserve SDK recovery state and stop native restart loops
Nov 6, 2026
62266a8
fix: recover verified vector checkpoints without consuming the pendin…
Nov 6, 2026
98c8723
fix: preserve disk caller admission and await checkpoint metadata con…
Nov 6, 2026
7343a8e
fix(capture): bound retained inbox reads and selected recovery
Nov 7, 2026
f8246e5
fix(capture): allow SDK queues to drain at capacity
Nov 7, 2026
59dbc72
fix(embedding): scope deferred source guards to selected keys
Nov 7, 2026
c8a7014
fix(capture): bound SDK handoff deadlines and advance embedding backlog
Nov 7, 2026
fb32895
fix(state): acknowledge large summary writes without value echo
Nov 8, 2026
152fa91
fix: preserve manual hold during storage and memory recovery
Nov 8, 2026
87a3d10
fix: complete retained embeddings and tune memory protection
Nov 10, 2026
746a3dd
fix: expose bounded retained vector proof diagnostics
Nov 10, 2026
29362ab
fix(health): recognize SDK reviews and legacy summary coverage
Nov 11, 2026
79fdcbb
fix(graph): shard retrieval index builds
Nov 11, 2026
f3579e4
chore(upstream): stage exact v0.9.33 fixes on isolated review branch
Daniel-Hong Nov 13, 2026
612f7d1
ci(personal): validate review PRs with unit tests and generated-clien…
Daniel-Hong Nov 13, 2026
796c2da
fix(upstream): preserve custom capture normalization while exposing v…
Daniel-Hong Nov 13, 2026
1cc3850
fix(build): pin yaml and json5 to bundled SDK audit exact versions
Daniel-Hong Nov 13, 2026
967f484
build(upstream): synchronize generated hook bundles with imported v0.…
Daniel-Hong Nov 13, 2026
8ecd207
build(hooks): retain executable mode on 8 regenerated hook scripts
Daniel-Hong Nov 13, 2026
08bb147
merge: preserve operating daniel.68 and personal capture proofs
Nov 13, 2026
c35903a
merge: integrate upstream 0.9.33 with verified Ubuntu State and captu…
Nov 14, 2026
73b86f3
test: make Ubuntu CI fixtures portable and deadline-aware
Nov 14, 2026
bee90e0
test: synchronize browser readiness and preserve publication deadline…
Nov 14, 2026
01d9a55
fix: preserve bounded SDK queue transitions on ext4
Nov 14, 2026
01d4f98
build: regenerate plugin bridge after candidate version synchronization
Nov 14, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
2 changes: 1 addition & 1 deletion .cursor-plugin/plugin.json
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
{
"name": "agentmemory",
"version": "0.9.32-daniel.55",
"version": "0.9.33-daniel.73",
"description": "Persistent memory for Cursor: auto-captures sessions, prompts, and tool use via hooks, recalls with hybrid BM25 + vector + graph search, and exposes 55 memory tools over MCP. Keyless by default, runs entirely on your machine.",
"author": {
"name": "Rohit Ghumare",
Expand Down
11 changes: 7 additions & 4 deletions .env.example
Original file line number Diff line number Diff line change
Expand Up @@ -130,13 +130,15 @@
# MAX_OBS_PER_SESSION=500 # Hard per-session cap: later observations of that session are refused; only capture-worker refusals are parked as dead letters (0 = no cap)
# 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.
# SUMMARIZE_CHUNK_CONCURRENCY=6 # Parallel chunk LLM calls during chunked summarize. Default 6 fits ~100-chunk sessions under iii's 180s function-invocation timeout at typical ~8s/call. High-throughput providers (Novita, DeepInfra, DeepSeek) commonly allow 100+ concurrent — bump this for very large imported sessions.
# AGENTMEMORY_SUMMARY_TIMEOUT_MS=120000 # Total time budget in whole milliseconds for one mem::summarize run, shared by every chunk call. Default 120000 (2 min). Raise it for slow or local LLMs (Ollama, LM Studio, vLLM) on large chunked sessions that fail with summary_timeout. Session-end and recovery summaries wait this budget plus 60 s; POST /agentmemory/summarize is still cut off by the 180 s HTTP worker timeout, but the summary keeps running and is saved.

# -----------------------------------------------------------------------------
# 5. Behaviour flags
# -----------------------------------------------------------------------------

# 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 # Opt in to Claude-style hook context injection. Pi/OMP use 1 in their launch environment for file enrichment; native prompt recall is independent.
# AGENTMEMORY_ENRICH_TIMEOUT_MS=2000 # PreToolUse hook wait (default 2000 ms). Set this in the hook process environment; setting it only in this server env file does not change the hook timeout.
# 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
Expand Down Expand Up @@ -168,6 +170,7 @@
# 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_CALL_TIMEOUT_MS=190000 # MCP shim and plugin bridge timeout for each proxied REST call, read from the MCP host environment or this file. Default 190000, a little above the 180 s worker invocation limit, so long calls such as memory_consolidate and memory_reflect end on the server first
# 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
Expand Down Expand Up @@ -215,10 +218,10 @@
# 8. iii engine pin
# -----------------------------------------------------------------------------
#
# 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
# agentmemory pins iii-engine and its iii-sdk dependencies to v0.25.1.
# The CLI downloads the engine into ~/.agentmemory/bin, including the Windows ZIP.
# AGENTMEMORY_III_VERSION=0.25.1 # Override only when deliberately matching a different engine
# AGENTMEMORY_LOCAL_DISPATCH=true # iii 0.25 runs calls to the worker's own functions in-process; set false for full engine traces

# Engine usage telemetry (iii). Off unless you set it; true opts in.
# III_TELEMETRY_ENABLED=false
Expand Down
2 changes: 2 additions & 0 deletions .gitattributes
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
# Unified diff context lines intentionally contain one leading space, including blank context.
deploy/state-worker/compat.patch whitespace=-blank-at-eol
2 changes: 1 addition & 1 deletion .github/workflows/ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ jobs:
os: [ubuntu-latest, macos-latest]
node-version: [20, 22, 24, 26]
steps:
- uses: actions/checkout@v6
- uses: actions/checkout@v7
with:
persist-credentials: false
- uses: actions/setup-node@v6
Expand Down
49 changes: 49 additions & 0 deletions .github/workflows/personal-pr-validation.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,49 @@
name: Personal fork PR checks

# Source-only validation for changes proposed against personal.
# No production credentials, service control, data migration or deployment.
on:
pull_request:
branches: [personal]
types: [opened, synchronize, reopened, ready_for_review]

permissions:
contents: read

concurrency:
group: personal-pr-checks-${{ github.event.pull_request.number }}
cancel-in-progress: true

jobs:
source:
name: Source, generated hooks and unit regressions
runs-on: ubuntu-latest
timeout-minutes: 35
steps:
- uses: actions/checkout@v6
with:
persist-credentials: false
- uses: actions/setup-node@v6
with:
node-version: 22
- name: Install dependencies
run: npm install --legacy-peer-deps --no-audit --no-fund
- name: Build
run: npm run build
- name: Generate explicit Codex compatibility artifacts
run: npm run build:codex-compat
- name: Reject stale checked-in hook bundles
run: git diff --exit-code -- plugin/scripts
- name: Verify skills and documented facts
run: |
npm run skills:check
npm run docs:check
- name: Eval TypeScript contract
run: npx tsc --noEmit --project tsconfig.eval.json
- name: Unit and non-integration regressions (isolated HOME)
shell: bash
run: |
test_home="$(mktemp -d "${RUNNER_TEMP}/agentmemory-pr-home.XXXXXX")"
trap 'rm -rf "$test_home"' EXIT
export HOME="$test_home"
npm test
4 changes: 2 additions & 2 deletions .github/workflows/publish.yml
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ jobs:
contents: read
id-token: write
steps:
- uses: actions/checkout@v6
- uses: actions/checkout@v7
with:
persist-credentials: false

Expand Down Expand Up @@ -56,7 +56,7 @@ jobs:
NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}

- if: always()
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
with:
name: npm-release-${{ github.run_id }}-${{ github.run_attempt }}
path: ${{ runner.temp }}/npm-release
Expand Down
22 changes: 13 additions & 9 deletions .github/workflows/release-gate.yml
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ jobs:
strategy:
fail-fast: false
matrix:
os: [ubuntu-latest, macos-latest]
os: [ubuntu-latest, ubuntu-24.04-arm, macos-latest, macos-15-intel]
steps:
- uses: actions/checkout@v6
with:
Expand All @@ -49,9 +49,9 @@ jobs:
shell: bash
run: |
engine_pin=$(node --import tsx --input-type=module -e 'import { III_PINNED_VERSION } from "./src/version.ts"; console.log(III_PINNED_VERSION)')
engine_bin="$HOME/.cache/agentmemory-release-gate/iii-${engine_pin}-linux-x64"
engine_bin="$HOME/.cache/agentmemory-release-gate/iii-${engine_pin}-linux-$(node -p process.arch)"
test -x "$engine_bin"
AGENTMEMORY_TEST_III="$engine_bin" AGENTMEMORY_REQUIRE_LIVE_TESTS=1 npx vitest run test/compose-supervision-integration.test.ts test/codex-live-daemon.test.ts
AGENTMEMORY_TEST_III="$engine_bin" AGENTMEMORY_REQUIRE_LIVE_TESTS=1 npx vitest run test/compose-supervision-integration.test.ts test/codex-live-daemon.test.ts test/compose-c1-integration.test.ts
- run: npm run test:startup:packaged -- --skip-build --out "${{ runner.temp }}/startup-gate"
- name: Test packaged vector recovery
shell: bash
Expand All @@ -74,9 +74,9 @@ jobs:
archives=("$RUNNER_TEMP"/graph-index-pack/*.tgz)
npm run test:graph-index:packaged -- --tarball "${archives[0]}" --out "$RUNNER_TEMP/graph-index-gate"
- if: always()
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
with:
name: release-gate-${{ matrix.os }}
name: release-gate-${{ matrix.os }}-${{ github.run_attempt }}
path: |
${{ runner.temp }}/release-gate
${{ runner.temp }}/startup-gate
Expand All @@ -88,9 +88,13 @@ jobs:
retention-days: 14

windows-smoke:
name: windows smoke
runs-on: windows-latest
name: windows smoke (${{ matrix.os }})
runs-on: ${{ matrix.os }}
timeout-minutes: 30
strategy:
fail-fast: false
matrix:
os: [windows-latest, windows-11-arm]
steps:
- uses: actions/checkout@v6
with:
Expand All @@ -117,9 +121,9 @@ jobs:
node scripts/release-gate/windows-smoke.mjs --tarball $tarballs[0].FullName --out "$env:RUNNER_TEMP\windows-smoke"
if ($LASTEXITCODE -ne 0) { exit $LASTEXITCODE }
- if: always()
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@v7
with:
name: windows-smoke
name: windows-smoke-${{ matrix.os }}-${{ github.run_attempt }}
path: ${{ runner.temp }}/windows-smoke
if-no-files-found: ignore
retention-days: 14
12 changes: 6 additions & 6 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@

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 0.25.0 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`)
- **Engine**: iii-sdk 0.25.1 with @iii-dev/helpers 0.25.1 (Worker Compose-managed iii-engine 0.25.1 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)
Expand Down Expand Up @@ -108,20 +108,20 @@ 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` (6,300+ 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` (7,000+ 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 (v0.9.32-daniel.55)
## Current Stats (v0.9.32-daniel.67)

- 55 MCP tools visible by default; `AGENTMEMORY_TOOLS=core` selects 8 essentials
- 160 REST endpoints
- 161 REST endpoints
- 6 MCP resources, 3 MCP prompts
- 12 hooks, 17 skills
- 353+ iii functions
- 6,300+ tests
- 355+ iii functions
- 7,000+ tests

## Personal fork verification isolation

Expand Down
Loading
Loading