Canary: internal users first

Your first line in any agent design is what the agent must not do. Most failures come from unclear scope, sloppy tool contracts and missing failure handling, not from a weak model.

Task: A new agent going live, with a fear of a big failure.

[Before you start, do three things]
1. Restate the goal in one sentence: who reads the output, and what decision it serves.
2. If I have not given you the following, ask me. Do not assume:
   - blast radius and user types
   - observable metrics
   - how to revert
   - canary duration
3. Say how you plan to do it (at most 5 steps) and wait for my go-ahead.

[Go through each of these]
- three batches: internal, one percent, all
- criteria per batch, including sampled quality not just error rate
- keep the old path switchable
- collect and triage failures same day

[Process]
1. Write the boundary: what it may do, must never do, and when to hand back to a human
2. Define each tool's inputs, outputs, failure semantics, idempotency
3. Write the flow as a state machine with exit conditions
4. Design failure paths: retry, degrade, abort, escalate
5. Give one testable success criterion

[Output]
Fixed shape: boundary, tool contracts, state machine, failure paths, criteria.

[Do not]
- No general-purpose agent that does everything
- Never let fuzzy prose stand in for a tool contract
- The agent must not decide whether approval is needed
- No 'do your best' instead of an abort condition