Ship and explain the research-critique coordinator

A shippable lab artifact includes: how to run it, the metric result (veto rate; unsupported claim count after merge; total step budget), the failure you can still reproduce (workers sharing mutable state, or coordinator ignoring critique veto), the security gate for critique worker receiving secrets from research scratchpad logs, and a rollback note. The user decision it supports remains: delegate research and critique to two bounded workers, then merge under a coordinator policy.

RUN_LIVE_EVALS=1 npm run test:live

Expected evidence: **Research returns two sourced claims; critique rejects one unsupported claim; synthesis keeps one.**. Store this beside the fixture version so scores remain meaningful after content changes in `multi-agent-in-code`.

State operating limits for the research-critique coordinator in plain language: fixture size, offline vs live dependencies, and what would require a new eval set. Shipping multi-agent-in-code is honest scoping, not maximal confidence language.

Fill this table in your notes for the research-critique coordinator:

• Proved on two workers with separate step budgets: …
• Unproved beyond the fixture: …
• Metric that blocks release: veto rate; unsupported claim count after merge; total step budget
• Failure still reproducible: workers sharing mutable state, or coordinator ignoring critique veto
• Security gate: critique worker receiving secrets from research scratchpad logs
• Rollback: …

Ship the narrative only when the unproved list is honest. Reviewers trust narrow claims that support delegate research and critique to two bounded workers, then merge under a coordinator policy more than maximal language that collapses under the first production oddity.

Hand your ship note to a peer and ask them to recreate a proved/unproved ship note with rollback without watching you type. If they cannot, your evidence is still tribal knowledge. Tighten the run command and the metric line until a stranger can validate the research-critique coordinator against two workers with separate step budgets.

At the mastery and shipping stage for multi-agent-in-code, the job is narrower than finishing a product demo. You are creating one progressive evidence piece about two workers with separate step budgets that later pages inherit without redefining success. Keep that fixture small enough to inspect by hand, keep outputs copy-pasteable as text, and refuse to narrate this baseline as if it were a production SLA: single agent without critique pass.

For this page specifically, success looks like a proved/unproved ship note with rollback while still centering the user decision to delegate research and critique to two bounded workers, then merge under a coordinator policy. If you cannot point to a file, command, or assertion that proves that for the research-critique coordinator, stay on this page instead of advancing.

Glossary: tool · Glossary: structured output · Cheatsheet: production ops signals

Previous