Spec Kitty Runtime Review

spec-kitty-runtime-review

Operate the Spec Kitty review workflow surface: load review context, claim a work package, read the generated review prompt, and issue an approve or reject transition.

When to Use This Skill

• Claim a completed work package for review
• Read the review prompt generated by the review workflow
• Issue an approval or rejection with structured feedback

Step 1: Load Review Context

spec-kitty charter context --action review --json

The returned text contains governance context. The review prompt (generated in Step 2) includes project-specific acceptance criteria and review guidance from doctrine — do not restate those rules here.

Step 2: Claim the Work Package

# Claim a specific WP (or omit WP## to auto-select from for_review lane)
spec-kitty agent action review WP## --agent <your-name>

This moves the WP from for_review to in_review and prints the path to a generated review prompt file. Read that path from the command output.

Step 3: Read the Review Prompt

cat <prompt-file-path>

The review prompt contains:

• Acceptance criteria for this specific WP
• Git diff commands with the correct base branch (use those, not hardcoded main)
• Dependency warnings if the WP has downstream dependents
• WP isolation rules
• Completion instructions (approve/reject commands)

Follow the review prompt. It is the source of truth for what to check and how to check it. The review criteria come from doctrine and the WP definition, not from this skill.

Step 3.5: Contract Round-Trip Check

If the mission ships a kitty-specs/<mission>/contracts/ directory, walk it before issuing a verdict. Contracts pin concrete examples (payloads, CLI invocations, schema fragments, error messages, allowed-value sets) that the implementation must round-trip verbatim. Vocabulary or shape drift here is invisible to spec-level review and surfaces late at mission-review or downstream consumption time.

1. Enumerate — ls kitty-specs/<mission>/contracts/. If the directory is absent, record a one-line note ("no contracts/ artifact") and skip to Step 4.
2. Filter — for each contract file, decide whether this WP touches the symbols, payloads, or commands it pins. Skipped files get a one-line note ("orthogonal to WP##").
3. Walk every concrete example — extract every YAML block, JSON snippet, CLI invocation, schema fragment, and error message in the in-scope contracts. Verify the implementation accepts inputs and produces outputs verbatim: singular vs plural, default values, enum members, exit codes, flag names — all of it.
4. SSOT cross-check — if a contract pins a closed set (allowed values, registered triggers, enum members), require one assertion that mirrors the runtime constant against its test-side or doc-side copy byte-for- byte. If the WP introduces such a constant without that mirror assertion, reject.
5. Verdict — any contradiction between the implementation and a contract example is a BLOCKER, cited as contract file:line plus impl symbol. Contracts are spec artifacts, not preferences.

Step 4: Issue Verdict

Take exactly one action — never "approve with conditions".

Approve (all acceptance criteria met)

spec-kitty agent tasks move-task WP## --to approved --note "Review passed: <summary>"

Reject (acceptance criteria not met)

Write structured feedback to a temp file, then move the WP back to planned:

spec-kitty agent tasks move-task WP## --to planned --force \
  --review-feedback-file <feedback-file-path>

Every blocking finding must map to a specific, verifiable remediation action.

Step 5: Check Downstream Impact

If you rejected and the WP has downstream dependents:

spec-kitty agent tasks status --mission <mission-slug>

Note dependent WPs and include a rebase warning in your feedback.

Review Precedence Rules

1. Acceptance criteria are the primary gate — a WP meeting all criteria passes even if the reviewer would have done it differently
2. The review prompt is the source of truth — it contains the specific checks, criteria, and doctrine context for this WP
3. One clear verdict per review — approve or reject, nothing in between
4. The reviewer does not implement fixes — feedback must be actionable by the original implementing agent