What is Sovereign AI?

Sovereign AI is an autonomous build executor that takes council verdicts and builds them without human intervention. It generates structured plans via Claude, writes production TypeScript, type-checks every step, auto-fixes errors with a retry loop, and deploys the result.

How does Sovereign handle build failures?

Sovereign runs an auto-fix loop with 2 retries per step. When a typecheck fails, it sends the errors and file contents back to Claude for diagnosis and correction. If all retries are exhausted, Sovereign reverts via git and flags the plan for human review.

What safety measures does Sovereign have?

Sovereign blocks destructive patterns including rm -rf, git push, and sudo. High-risk plans are flagged for human review instead of being executed automatically. Failed typechecks trigger a git revert. SENTINEL monitors all builds for anomalies and budget overruns.

Can Sovereign build full applications?

Yes. Sovereign generates complete build plans with file paths, component trees, API endpoints, and database schemas. Each step produces complete file contents — not patches or diffs — so every build artifact is fully formed and ready for production.

How does Sovereign integrate with the council?

Council sessions produce verdicts containing build specs. Sovereign's verdict parser extracts those specs, queues them in the build queue, and auto-approves specs estimated under 4 hours of effort. Larger or high-risk specs are held for manual approval before execution begins.

SOVEREIGN

Autonomous build executor — submit jobs, track progress, manage the pipeline

NEW BUILD REQUEST

Cmd+Enter to queue · Shift+Cmd+Enter to queue & build

Loading pipeline...