FastClaw / Operations, Maintenance and Governance
deploy
FastClaw can be run locally on a single machine, or can be used for production deployment using Postgres, Redis, object storage, and Docker/E2B/Boxlite sandboxes.
local deployment
The minimal deployment is a fastclaw process and SQLite. Suitable for personal development, demo, internal experiments and stand-alone Agent platform.
fastclaw
fastclaw daemon start
fastclaw daemon statusProduction components
When deploying with multiple users or multiple replicas, the README and source code point to Postgres, Redis, object storage, and external sandbox backends. Postgres is the source of truth, Redis is the bus/leases, and the object store synchronizes files and skills.
| components | when needed |
|---|---|
| Postgres | Multiple copies, stable backup, production data |
| Redis | Cross-process buses, leases, real-time collaboration capabilities |
| S3/R2 Object Storage | File, skills, artifact synchronization |
| Docker/E2B/Boxlite | Isolated execution, project preview, code running |
Key environment variables
FASTCLAW_* variables cover startup configurations such as home, port, bind, storage, redis, sandbox, object store, log, etc. Long-term provider/channel configuration is still recommended into DB.
- FASTCLAW_HOME: Data root directory.
- FASTCLAW_STORAGE_DSN: Postgres connection string.
- FASTCLAW_REDIS_*: Redis configuration.
- FASTCLAW_SANDBOX_*: Sandbox backend configuration.
- FASTCLAW_OBJECTSTORE_*: Object storage configuration.