| name | final-report |
| description | Final response contract for ultracode-high and high-rigor subagent workflows. Use to synthesize work into a relayable result with changes, verification, final-gate improvements, risks, and next actions. |
| model | claude-opus-4-8 |
| effort | high |
Final Report
The final answer is the product. Lead with the answer, not the process.
Fleet context
This skill is part of ultracode-high. For the full orchestration map, tool
catalog, agent roster, and card schemas, read
../ultracode-high/references/runtime-context.md when available. Your output is
the only thing the user can safely act on, so carry verification, caveats, and
attestation labels forward.
Shape
RESULT: <answer, recommendation, or deliverable summary>
CHANGES: <files/artifacts changed or produced>
VERIFICATION: <checks passed, failed, skipped, gate types, with evidence>
IMPROVED BY FINAL GATE: <specific changes made after critique>
RISKS: <decision-changing assumptions, caveats, or unknowns>
NEXT: <only if human action is required>
Gate-results table
For S2+ work, render VERIFICATION as a table so measured and judged checks are
distinguishable at a glance, with every attestation row flagged:
| check | gate_type | status | evidence |
|---|
| unit tests | command | passed | npm test exit 0 |
| metric reconciles to source of truth | command | passed | recon query diff = 0 |
| narrative reads clearly for the audience | attest (flagged) | passed | editor judgment |
A high attestation ratio on code/data work is a signal that the task was
under-specified, not that it was thoroughly verified.
Rules
- Do not claim verification that did not run.
- Surface the command-vs-attest gate mix for S2+ work. Attestation is allowed,
but it must not masquerade as mechanical verification.
- Distinguish result from remaining risk.
- Include file paths when artifacts were created or edited.
- Keep raw logs out unless the user asked for them.
- For blocked work, include the smallest concrete question or artifact needed.
