@clawtex/mcp-server

Structured memory for AI agents. Gives your agent persistent state, event history, and AI-extracted lessons across sessions.

Setup

1. Get an API key

Sign up at clawtex.io and create an agent. Your API key starts with tkr_.

2. Add to your Claude config

Claude Code — edit ~/.claude/settings.json:

{
  "mcpServers": {
    "clawtex": {
      "command": "npx",
      "args": ["-y", "@clawtex/mcp-server"],
      "env": {
        "CLAWTEX_API_KEY": "tkr_your_key_here"
      }
    }
  }
}

Claude Desktop — edit the config file at:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the same clawtex block inside mcpServers.

3. Restart Claude

Start a new session. Clawtex loads your state and lessons automatically.

4. Verify it works

Ask Claude: "What Clawtex tools do you have?"

It should list 15 tools including clawtex_bootstrap, clawtex_log_event, and clawtex_update_state. Your dashboard at clawtex.io will show the agent as "Active".

Tools

| Tool | Description | |------|-------------| | clawtex_bootstrap | Load full context (state + lessons + events + rules) | | clawtex_get_state | Read state entities | | clawtex_update_state | Create or update a state entity | | clawtex_log_event | Log a decision, milestone, blocker, task, or deployment | | clawtex_get_events | Query event history | | clawtex_get_lessons | Get active lessons | | clawtex_extract_lessons | Trigger AI lesson extraction from events | | clawtex_create_lesson | Manually create a lesson | | clawtex_approve_lesson | Approve a pending lesson | | clawtex_reject_lesson | Reject a pending lesson | | clawtex_signal_lesson | Track lesson effectiveness | | clawtex_start_session | Start a tracked work session | | clawtex_end_session | End a session with summary | | clawtex_get_sessions | View recent sessions | | clawtex_search | Search across all memory |

Documentation

Full docs at clawtex.io/docs

License

Business Source License 1.1