Operator guide · three minute read
The two commands an operator runs most, for the security team running Interpose day to day. scan shows what an agent could reach before you enforce. release restores an agent that was escalated.
$ interpose scan walking payments-api · 2,318 files · read only high .env.production credential pattern, root high config/aws.json cloud config, key material medium scripts/deploy.sh shell, touches infra medium src/auth/keys.ts key material by name low src/** 1,904 files start policy from the high and medium rows writes nothing unless you pass --save
interpose scan in a project root lists every file an agent could touch, scored by how sensitive each one looks by filename and path pattern. It does not read file contents, so it is a blast radius map, not a found secrets claim.
--save. Add --path=../other-repo --format=json to point it elsewhere or pipe into tooling.--path or --force, so it never crawls a home directory by accident. Symlinks are skipped.$ interpose release --agent=payments-api-4f2 agent was quarantined after 5 violations last stop: shell egress to paste.lol, 14:02:38 restored to normal release written to the signed trail, entry 9 proposed: allow egress to api.stripe.com, this role, 7 days waiting on security to approve or reject
Repeated violations escalate an agent: normal, then restricted (the agent keeps working and every action is flagged for review), then quarantined (the agent is stopped until an operator releases it), at thresholds set in the policy. interpose release --agent=<id> returns it to normal. Run it from a separate terminal if the agent's own shell is blocked, so you are never locked out of the recovery path.
interpose --verify-audit--agent=<id> to scope it.interpose --show-auditClaude Code is the only live integration. scan scores by filename and path pattern, never by reading contents. This page reflects the present shipped state and moves as the product ships.
The full command reference, the policy schema, and the trail format ship with the pilot. For the architecture, read the technical brief. Questions: james@tuent.ai or charlie@tuent.ai.