SWARM-OpenClaw Bridge¶

Run SWARM as a production service with OpenClaw integration.

Overview¶

OpenClaw provides secure multi-agent orchestration. SWARM-OpenClaw enables:

REST API for scenario execution
Job queue for async simulation runs
Skill integration for OpenClaw workflows

Installation¶

pip install swarm-openclaw

Quick Start¶

Start the Service¶

swarm-service start --port 8000

Run a Scenario¶

curl -X POST http://localhost:8000/runs \
  -H "Content-Type: application/json" \
  -d '{"scenario": "baseline", "seed": 42}'

Get Results¶

curl http://localhost:8000/runs/{job_id}/metrics

API Reference¶

POST /runs¶

Create a new simulation run.

Request:

{
  "scenario": "baseline",
  "seed": 42,
  "epochs": 20,
  "steps_per_epoch": 15
}

Response:

{
  "job_id": "abc123",
  "status": "queued"
}

GET /runs/{job_id}¶

Get run status.

Response:

{
  "job_id": "abc123",
  "status": "completed",
  "epochs_completed": 20
}

GET /runs/{job_id}/metrics¶

Get run metrics.

Response:

{
  "toxicity_rate": 0.15,
  "quality_gap": 0.23,
  "total_welfare": 145.7
}

OpenClaw Skill¶

Use SWARM from OpenClaw workflows:

from openclaw import Skill

swarm_skill = Skill("swarm")

# Run scenario
result = await swarm_skill.run_scenario(
    scenario="baseline",
    seed=42
)

# Get metrics
metrics = await swarm_skill.get_metrics(result.job_id)

Docker Deployment¶

FROM python:3.11-slim
RUN pip install swarm-openclaw
EXPOSE 8000
CMD ["swarm-service", "start", "--host", "0.0.0.0"]

docker build -t swarm-service .
docker run -p 8000:8000 swarm-service

Status¶

In Development - Service layer functional, OpenClaw skill in progress.