skills.rest
Priivacy-ai
Priivacy-aiOFFICIAL
Explore all skills by this author →
GitHub

spk-run-verdict-capture

Records work-package review verdicts through the deterministic event-log seam in Spec Kitty.

1.6k|156|Updated Oct 9, 2025
Search Tags:#agent workflow#CLI#Event Sourcing#work package#spec-kitty#Code review#verdict capture
One-click install
npx skills add https://github.com/Priivacy-ai/spec-kitty --skill spk-run-verdict-capture
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: spk-run-verdict-capture
Source: https://github.com/Priivacy-ai/spec-kitty/tree/main/src/charter/offering/skills/spk-run-verdict-capture
Command: npx skills add https://github.com/Priivacy-ai/spec-kitty --skill spk-run-verdict-capture

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

When multiple AI agent harnesses (Claude, Codex, Gemini, Copilot, and others) review work packages, verdicts can diverge if each harness records approve/reject outcomes differently. This Skill defines the single deterministic seam for capturing review verdicts so every harness records them identically in the mission's event log.

Core Features & Use Cases

  • Deterministic verdict capture: Routes all approve/reject outcomes through the review_result event in status.events.jsonl, making the verdict append-only and identical across agents.
  • CLI-driven commands: Provides exact spec-kitty agent tasks move-task and spec-kitty agent status emit commands for approving, rejecting with structured feedback, or marking a WP done with a review triple (reviewer, verdict, reference).
  • Verdict validation: Enforces the canonical verdict vocabulary from specify_cli.status.event_verdicts() plus proof-event extras, rejecting unknown verdict strings at the seam.
  • Use Case: A reviewer agent finishes reviewing WP-03 and needs to reject it with actionable feedback; this Skill directs it to run move-task WP03 --to planned --review-feedback-file review-feedback-1.md so the rejection is recorded as an authoritative event rather than edited prose.

Quick Start

Ask the agent to record the review verdict for a work package using the spec-kitty CLI, for example approving WP-01 with a note or rejecting it with a review feedback file.

Frequently Asked Questions about spk-run-verdict-capture

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I approve a work package in Spec Kitty?▼

Approve a work package by running spec-kitty agent tasks move-task <WP> --to approved --mission <slug> --note "<reason>". This emits a review_result event into the mission's status.events.jsonl, which is the authoritative record of the verdict.

How do I reject a work package with feedback?▼

Reject a work package with spec-kitty agent tasks move-task <WP> --to planned --review-feedback-file <path> --mission <slug>. The rejection must include a feedback file with concrete, actionable comments; an empty rejection is not considered a valid review.

Can I edit the review-cycle markdown file to change a verdict?▼

No. The review-cycle-N.md render is non-authoritative prose and must never be hand-edited to change a verdict. The sole source of truth is the review_result event in status.events.jsonl, written only through the spec-kitty CLI commands.

What verdict values are allowed when marking a work package done?▼

Verdicts are validated against the canonical bridge specify_cli.status.event_verdicts() plus the proof-event extras commented, rejected, and unknown. Moving a WP to done requires a review triple containing reviewer, verdict, and reference; unknown verdict strings are rejected at the seam.

Why do all agent harnesses need to use the same verdict capture method?▼

Because the verdict is stored as an append-only event, using one CLI seam guarantees the recorded outcome is deterministic and identical across Codex, Claude, Gemini, Copilot, and other harnesses. Divergent recording methods would break merge preflight and status reporting.

Related Skills

Explore More Software Engineering Skills→
zivtechzivtech

webchick-review

Evaluate patches or MRs using webchick's review criteria and verdict framework.

Official
Intermediate
hw-native-syshw-native-sys
34

clean-branches

Remove local and fork branches merged into main, including squash merges.

Official
Intermediate
NodeJSmithNodeJSmith
1

mine.orchestrate

Automate development work packages through implementation, review, and quality assessment loops.

Community
Advanced
chdalychdaly

history-hygiene

Record confirmed final outcomes to history.md and update reversed decisions.

Community
Intermediate