ClawXMemory
https://github.com/OpenBMB/ClawXMemory- Category
- AI Agents
- Rank
- No. 1549Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- openbmb
- GitHub
- 49 stars
- Latest release
- v0.1.5
- Date
About
Multi-level memory plugin for OpenClaw that adds long-term context across agent sessions.
What it does
ClawXMemory converts conversations into inspectable Markdown records for user profiles, projects, and collaboration feedback. Background indexing extracts those records, while Dream consolidates or removes superseded material. At answer time, a model decides whether memory is needed, chooses one project, and loads selected evidence. SQLite holds operational state rather than durable memory.
Why it's ranked here
The architecture makes a strong, understandable trade: durable knowledge stays readable on disk, while retrieval limits prompt growth by selecting relevant files. The dashboard and traces add useful visibility into indexing, recall, and consolidation. Its value is narrower than a general memory service because it is tied to OpenClaw and model-guided selection of one formal project.
What's good
Markdown remains the durable source of truth, so people can inspect and migrate important context without decoding a proprietary store. Automatic extraction reduces manual upkeep. Recall first checks whether memory is relevant, then narrows project and file choices. The local dashboard exposes records, runtime health, and traces, while explicit tools support searching, browsing, reading, flushing, and consolidation.
Tradeoffs
OpenClaw is a prerequisite, and installation changes memory ownership by disabling native memory features. Project recall selects one formal project, so ambiguous requests may require clarification. Model decisions govern routing and evidence selection. Background indexing and Dream can rewrite, merge, or delete superseded records. Uninstalling may also leave an extension directory that requires manual removal.
How to use it well
Use it when an OpenClaw agent repeatedly revisits long-running projects, user preferences, decisions, and delivery rules. Let automatic indexing capture normal work, inspect traces when recall looks wrong, and run manual flushing when new context must become searchable sooner. It does not replace a general database, cross-platform memory API, or broad search across several projects in one answer.
Technical notes+
clawxmemory/src/index.ts registers the memory prompt section, runtime adapter, agent tools, hooks, and service through the OpenClaw plugin SDK. clawxmemory/src/tools.ts defines search, overview, list, exact-read, flush, and Dream tools with TypeBox schemas. clawxmemory/src/config.ts defaults durable files beneath the local data directory, SQLite to memory.sqlite, and the dashboard to 127.0.0.1:39393/clawxmemory. clawxmemory/src/runtime.ts applies a managed boundary that selects this plugin for the memory slot and disables native memory search, session memory, and compaction flushing. clawxmemory/src/ui-server.ts serves the local HTTP interface and exposes controls for indexing, Dream, memory actions, import, export, and clearing. clawxmemory/vitest.config.ts configures Node-based Vitest discovery under tests/**/*.test.ts.
Observed
- License
- MIT
- Primary language
- TypeScript
- Packaging
- Published as the global npm package openbmb-clawxmemory, with ClawHub installation also documented
- Interface
- OpenClaw memory-slot plugin with registered agent tools and a local HTTP dashboard
- Storage
- Durable memory uses local Markdown files; SQLite stores runtime control-plane state, captured sessions, settings, and traces
- Development tooling
- Vitest is configured for Node tests, and repository scripts cover relinking, reloading, UI asset copying, and uninstalling
Read from README.md, docs/README_zh.md, clawxmemory/vitest.config.ts, clawxmemory/src/index.ts, clawxmemory/src/hooks.ts, clawxmemory/src/tools.ts, clawxmemory/src/config.ts, clawxmemory/src/runtime.ts, clawxmemory/ui-source/app.js, clawxmemory/src/ui-server.ts, clawxmemory/scripts/relink.mjs, clawxmemory/scripts/reload.mjs, clawxmemory/scripts/copy-ui.mjs, clawxmemory/src/message-utils.ts, clawxmemory/scripts/uninstall.mjs.
What it can do
Store agent conversation history across sessions
Agent dialogue and interaction data → Persistent memory records
Retrieve relevant context from previous sessions
Current conversation or query → Related historical context and information
Maintain long-term memory persistence
Agent session data → Continuously updated memory database
Enable context-aware responses using historical data
User queries and stored memory context → Enhanced agent responses with historical awareness
Organize memory into hierarchical levels
Raw conversation and interaction data → Structured multi-level memory storage
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.