An easy-to-use dashboard, CLI, and FastAPI backend for managing OpenClaw agents built for data teams. Spin up agents that understand your pipelines, query your warehouses, generate dbt models, review SQL, and more — from the web UI or the terminal. Built for data engineers and analysts who want AI assistance without leaving their stack.
Overview of your agents, system health, and quick actions.
Talk to any agent directly. Switch agents, view connection status, and review conversation history.
Set up agents with custom models, skills, and linked sub-agents.
Browse, enable, and install skills for your agents — databases, APIs, data tools, and more.
Track token usage and estimated costs across all agents and sessions.
The fastest way to get running:
# Copy env and configure
cp .env.example .env
# Edit .env with your OpenClaw gateway token
# Start everything
docker compose up --buildThis starts:
- API at http://localhost:8000 (FastAPI + OpenClaw gateway)
- Web UI at http://localhost:3000 (Next.js)
Data is persisted in Docker volumes (clawdata-db, clawdata-userdata).
To stop: docker compose down (add -v to also remove volumes/data).
# Create venv and install
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
# Copy env and configure
cp .env.example .env
# Edit .env with your OpenClaw gateway token
# Run migrations
alembic upgrade head
# Start the server
uvicorn app.main:app --reload --port 8000cd web
npm install
npm run dev
# Open http://localhost:3000ClawData includes an interactive CLI for managing agents and skills without the web UI:
# Launch the interactive menu
clawdata
# Or jump straight to a section
clawdata skills # Toggle skills on/off per agent (↑↓ navigate, Space toggle, Enter confirm)
clawdata agents # Browse, create, and configure agents
clawdata gateway # Gateway health, logs, models, and providers
clawdata templates # Browse project templates
clawdata status # Quick system overview
clawdata setup # Guided setup wizard
clawdata tui # Launch the OpenClaw TUISkill management uses an interactive checklist — arrow keys to move, space bar to toggle, enter to apply:
Skills for 🦞 main (↑↓ navigate, Space toggle, Enter confirm)
> [x] Azure
[ ] BigQuery
[x] Data Analysis
[x] Databricks Skill
[ ] dbt
[ ] DuckDB
...
All commands also have non-interactive variants for scripting:
clawdata skills list --project
clawdata skills deploy dbt --agent main
clawdata agents list --gateway
clawdata gateway healthOnce running, visit:
- Swagger UI: http://localhost:8000/docs
- ReDoc: http://localhost:8000/redoc
See PLAN.md for the full implementation plan.
| Directory | Purpose |
|---|---|
app/ |
FastAPI application (routers, services, adapters) |
app/cli/ |
Interactive CLI (clawdata command) |
templates/ |
Jinja2 reference templates (dbt, airflow, sql) |
skills/ |
Agent skill definitions (SKILL.md markdown) |
userdata/ |
Agent workspaces (head agent + sub-agents) |
migrations/ |
Alembic database migrations |
tests/ |
Test suite |
web/ |
Next.js 16 frontend (App Router, shadcn/ui) |




