Forge · run real work

Scope checks before execution

SourceA is an AI execution platform powered by Forge — real builds, automations, and agent workflows, with a verifiable receipt on every run. Forge enforces one job, safe commands, and your standing rules — before any agent spends a token.

How it works under the hood

Standing constraints (injected every time)

NO rebuild / redeploy / start over if it already exists — wire and fix.
One job only — no bundled unrelated work.
Do not break anything already working (TLS / live state).
Verdicts from receipts only — no build theater.

Policy detects

Destructive shell patterns · multi-job inputs · missing founder decisions (naming, paths, provider choice). Forge surfaces warnings before you waste a Cursor turn.

Law on disk

SSOT v3: brain-os/ssot/SOURCEA_LLM_AGENT_OPERATING_LAW_SSOT_v3.md — Forge is the outbound machine that applies this law to what you send agents.