Give Your OpenClaw Agent Long-Term Memory

The Awareness plugin gives your OpenClaw agent persistent, structured memory across sessions, IDEs, and your team.

Run a local memory daemon — no cloud account needed. The plugin auto-detects the daemon on localhost:37800.

One-Click Install Manual SetupFull Documentation
Start Local Daemon
npx @awareness-sdk/local start
No account neededNo API key neededData stays on your machine

Want cloud sync later? Run: npx @awareness-sdk/setup --cloud

One-Click Install - Let OpenClaw Do It

Copy this prompt and paste it to your OpenClaw agent. It will install and configure the Awareness Memory plugin automatically.

Help me install the Awareness Memory plugin for OpenClaw in local-first mode. Follow these steps exactly:

1. Start the local memory daemon (if not already running):
   npx @awareness-sdk/local start

2. Install via one of these methods:
   # Skill (recommended):
   openclaw skills install awareness-memory
   # Or via ClawHub CLI:
   npx clawhub@latest install awareness-memory
   # Or as npm plugin (full integration — requires --dangerously-force-unsafe-install
   # to bypass OpenClaw 2026.4+'s plugin scanner; child_process + env-read are both
   # legitimate features, see https://github.com/openclaw/openclaw/issues/11030):
   openclaw plugins install @awareness-sdk/openclaw-memory@latest --force --dangerously-force-unsafe-install

3. Verify by running: openclaw skills list (or openclaw plugins list)
   Confirm that "openclaw-memory" (or "awareness-memory") shows as loaded.
   No apiKey or memoryId needed — the plugin/skill auto-detects the local daemon.

If any step fails, show me the error and suggest a fix.

Local mode — no API key or Memory ID needed. Copy and paste directly.

Why Awareness for OpenClaw?

Cross-Session Memory

Your agent remembers past decisions, code changes, and architecture across sessions - no more repeating context.

Structured Knowledge

Auto-extracts KnowledgeCards, Risks, and ActionItems using YOUR LLM - zero extra inference cost.

Semantic Recall

Semantic + keyword hybrid search finds the most relevant past context, not just keyword matches.

Multi-User & Multi-Role

Team projects with per-user memory isolation. Each role sees its own relevant knowledge.

Knowledge Evolution

Conflict detection auto-supersedes outdated decisions. Your knowledge base stays current.

Cross-IDE Portability

Same memory works in Cursor, VS Code, Claude Code, or any MCP-compatible IDE.

Quick Setup

Step 0: Start local daemon
npx @awareness-sdk/local start
Step 1: Install the plugin

Plugin (full integration):

openclaw skills install awareness-memory

Or Skill (via ClawHub):

npx clawhub@latest install awareness-memory
Step 2: Configure in openclaw.json
{
  "plugins": {
    "slots": {
      "memory": "openclaw-memory"
    },
    "entries": {
      "openclaw-memory": {
        "enabled": true,
        "config": {
          "agentRole": "builder_agent",
          "autoRecall": true,
          "autoCapture": true,
          "recallLimit": 8
        }
      }
    }
  }
}

Local mode config — no credentials needed. Ready to use.

Step 3: Verify
openclaw plugins list

You should see openclaw-memory listed as an active plugin.

Feature Comparison

Capabilitymemory-coreAwareness
Cross-session recall
Structured knowledge extraction—
Conflict detection & supersede—
Task tracking (ActionItems)—
Risk identification—
Multi-user isolation—
Knowledge graph (entities)—
Daily narratives—
Cross-IDE portability
Memory marketplace—
Verified A Grade
100/100

End-to-end quality verified with DashScope Qwen-Plus - 4 KnowledgeCards, 2 Risks, 2 ActionItems extracted with complete summaries and correct categories.

Ready to give your agent memory?

Set up in under 2 minutes. Works with any LLM provider your OpenClaw agent already uses.

Read the Docs MCP Server Setup