Languages: English | ็ฎไฝไธญๆ
Give your AI agent persistent memory. One command. No account. Works offline.
Awareness Local is a local-first MCP memory server for AI coding agents. It gives Cursor, Claude Code, Copilot, Cline, and other MCP IDEs persistent memory, hybrid semantic + keyword retrieval, and reusable knowledge cards for long-running software projects.
It runs a lightweight daemon on your machine, stores memory as Markdown, indexes recall with SQLite FTS5 + embeddings, and keeps your AI workflow fast, explainable, and offline-ready.
npx @awareness.market/setup
That's it. Your AI agent now remembers everything across sessions.
AI coding agents lose context between sessions. Awareness Local provides cross-session memory recall so agents can continue work without re-explaining architecture, past decisions, pending tasks, and implementation constraints.
- Persistent memory for AI coding agents
- Local-first MCP server with offline support
- Hybrid retrieval (keyword + semantic)
- Knowledge card extraction for decisions, solutions, and risks
npx @awareness.market/setup
Then open your IDE and start coding. Awareness tools become available for recall, record, and session initialization.
- Long-running codebase migrations across many sessions
- Team handoffs where AI should remember prior implementation context
- Personal coding workflows that need durable preferences and conventions
- Multi-agent setups that share decision history and task memory
Yes. Local mode works fully offline with memory stored on your machine.
Memory is stored as Markdown in .awareness/
, with a local SQLite index for retrieval.
No. Cloud sync is optional and can be enabled later.
Any MCP-compatible IDE, including Cursor, Claude Code, Copilot, Cline, Windsurf, and others.
Evaluated on ** LongMemEval** โ the industry standard benchmark for long-term conversational memory. 500 human-curated questions across 5 core capabilities.
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ โ
โ Awareness Memory โ LongMemEval Benchmark Results โ
โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ
โ Benchmark: LongMemEval (ICLR 2025) โ
โ Dataset: 500 human-curated questions โ
โ Variant: LongMemEval_S (~115k tokens per question) โ
โ โ
โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ โ โ
โ โ Recall@1 80.2% (401 / 500) โ โ
โ โ Recall@3 92.8% (464 / 500) โ โ
โ โ Recall@5 96.0% (480 / 500) โ PRIMARY โ โ
โ โ Recall@10 98.6% (493 / 500) โ โ
โ โ โ โ
โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ
โ Method: Hybrid RRF (BM25 + vector, daemon pipeline) โ
โ Embedding: multilingual-e5-small (production model) โ
โ LLM Calls: 0 (pure retrieval, no generation cost) โ
โ Hardware: Apple M1, 8GB RAM โ 35 min total โ
โ โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Long-Term Memory Retrieval โ R@5 Leaderboard โ
โ LongMemEval (ICLR 2025, 500 questions) โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฌโโโโโโโโโโโโฌโโโโโโโโโโโโโโโโค
โ System โ R@5 โ Note โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโผโโโโโโโโโโโโผโโโโโโโโโโโโโโโโค
โ MemPalace (ChromaDB raw) โ 96.6% โ R@5 only * โ
โ โ
Awareness Memory (Hybrid) โ 96.0% โ Hybrid RRF โ
โ OMEGA โ 95.4% โ QA Accuracy โ
โ Mastra (GPT-5-mini) โ 94.9% โ QA Accuracy โ
โ Mastra (GPT-4o) โ 84.2% โ QA Accuracy โ
โ Supermemory โ 81.6% โ QA Accuracy โ
โ Zep / Graphiti โ 71.2% โ QA Accuracy โ
โ GPT-4o (full context) โ 60.6% โ QA Accuracy โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโดโโโโโโโโโโโโโโโโค
โ * MemPalace 96.6% is Recall@5 only, not QA Accuracy. โ
โ Palace hierarchy was NOT used in the evaluation. โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Awareness Memory โ R@5 by Question Type โ
โ โ
โ knowledge-update โโโโโโโโโโโโโโโโโโโโโโโโโโโ 98.7% โ
โ multi-session โโโโโโโโโโโโโโโโโโโโโโโโโโโโ 99.2%โ
โ single-session-asst โโโโโโโโโโโโโโโโโโโโโโโโโโโโ 98.2%โ
โ temporal-reasoning โโโโโโโโโโโโโโโโโโโโโโโโโโโ 93.2%โ
โ single-session-user โโโโโโโโโโโโโโโโโโโโโโโโโโ 92.9%โ
โ single-session-pref โโโโโโโโโโโโโโโโโโโโโโโโโโ 90.0%โ
โ โ
โ Overall โโโโโโโโโโโโโโโโโโโโโโโโโโโ 96.0%โ
โ โ
โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ Ablation Study โ โ
โ โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ โ
โ โ Vector-only: 92.6% โโโโโโโโโโโโโโโโ โ โ
โ โ BM25-only: 91.4% โโโโโโโโโโโโโโโโ โ โ
โ โ Hybrid RRF: 95.6% โโโโโโโโโโโโโโโโ โ
โ โ
โ โ (2026-04 harness run) โ โ
โ โ Hybrid = +3% over any single method โ โ
โ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โ
โ โ
โ arxiv.org/abs/2410.10813 awareness.market โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Zero LLM calls on retrieval (daemon path). Reproducible benchmark scripts โ
Before: Every session starts from scratch. You re-explain the codebase, re-justify decisions, watch the agent redo work.
After: Your agent says "I remember you were migrating from MySQL to PostgreSQL. Last session you completed the schema changes and had 2 TODOs remaining..."
Session 1 Session 2
โโโโโโโโโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Agent: "What database?" โ โ Agent: "I remember we โ
โ You: "PostgreSQL..." โ โ chose PostgreSQL for โ
โ Agent: "What framework?"โ โ โ JSON support. You had โ
โ You: "FastAPI..." โ โ 2 TODOs left. Let me โ
โ (repeat every session) โ โ continue from there." โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโโโโโโโโ
| IDE | Auto-detected | Plugin |
|---|---|---|
| Claude Code | ||
| โ | ||
awareness-memory |
CursorWindsurfOpenClaw@awareness.market/openclaw-memory
ClineGitHub CopilotCodex CLIKiroTraeZedJetBrains (Junie)AugmentAntiGravity (Jules)
Your IDE / AI Agent
โ
โ MCP Protocol (localhost:37800)
โผ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Awareness Local Daemon โ
โ โ
โ Markdown files โ Human-readable, git-friendly
โ SQLite FTS5 โ Fast keyword search
โ Local embedding โ Semantic search (optional: npm i @huggingface/transformers)
โ Knowledge cards โ Auto-extracted decisions, solutions, risks
โ Web Dashboard โ http://localhost:37800/
โ โ
โ Cloud sync (optional) โ
โ โ One-click device-auth โ
โ โ Bidirectional sync โ
โ โ Semantic vector search โ
โ โ Team collaboration โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
All memories stored as Markdown files in .awareness/
โ human-readable, editable, git-friendly:
.awareness/
โโโ memories/
โ โโโ 2026-03-22_decided-to-use-postgresql.md
โ โโโ 2026-03-22_fixed-auth-bug.md
โ โโโ ...
โโโ knowledge/
โ โโโ decisions/postgresql-over-mysql.md
โ โโโ solutions/auth-token-refresh.md
โโโ tasks/
โ โโโ open/implement-rate-limiting.md
โโโ index.db (search index, auto-rebuilt)
| Tool | What it does |
|---|---|
awareness_init |
|
| Load session context โ recent knowledge, tasks, rules | |
awareness_recall |
|
| Search memories โ progressive disclosure (summary โ full) | |
awareness_record |
|
| Save decisions, code changes, insights โ with knowledge extraction | |
awareness_lookup |
|
| Fast lookup โ tasks, knowledge cards, session history, risks | |
awareness_get_agent_prompt |
|
| Get agent-specific prompts for multi-agent setups |
Instead of dumping everything into context, Awareness uses a two-phase recall:
Phase 1: awareness_recall(query, detail="summary")
โ Lightweight index (~80 tokens each): title + summary + score
โ Agent reviews and picks what's relevant
Phase 2: awareness_recall(detail="full", ids=[...])
โ Complete content for selected items only
โ No truncation, no wasted tokens
Visit http://localhost:37800/
to browse memories, knowledge cards, tasks, and manage cloud sync.
Connect to Awareness Cloud for:
- Semantic vector search (100+ languages)
- Cross-device real-time sync
- Team collaboration
- Memory marketplace
npx @awareness.market/setup --cloud
Awareness Local is part of the Awareness ecosystem:
| Package | For | Install |
|---|---|---|
| Local daemon + MCP server | npx @awareness.market/setup |
|
wrap_openai() / wrap_anthropic() interceptors |
||
pip install awareness-memory-cloud |
||
wrapOpenAI() / wrapAnthropic() interceptors |
||
npm i @awareness-sdk/memory-cloud |
||
| Auto-recall + auto-capture | openclaw plugins install @awareness.market/openclaw-memory |
|
| Skills + hooks | /plugin marketplace add everest-an/Awareness-SDK โ /plugin install awareness-memory@awareness |
|
| One-command setup for 13+ IDEs | npx @awareness.market/setup |
Full SDK docs: awareness.market/docs
- Node.js 18+
- Any MCP-compatible IDE
No Python, no Docker, no cloud account needed.
If Awareness Local saves you from re-explaining your codebase to your AI agent, give it a โญ โ it helps more developers discover the project and pushes it toward GitHub Trending.
MIT
IDE Support: Cursor, Windsurf, Trae, Zed, VS Code, JetBrains. Compatible with: OpenClaw, AutoGPT, LangChain, MetaGPT. Key Technology: OMP (Open Memory Protocol), LatentMAS, Shared Thought Space, One-click Deployment. Focus: Solving AI "Lobster Memory" (Long-term memory loss), Automating complex workflows, Simplifying Agent setup.