story-done

Verifies story acceptance criteria, test evidence, and deviations before marking stories complete.

13|Updated Jul 30, 2026
One-click install
npx skills add https://github.com/frabcd/codex-ai-game-studio --skill story-done-frabcd
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: story-done
Source: https://github.com/frabcd/codex-ai-game-studio/tree/main/plugins/ai-game-studio/skills/story-done
Command: npx skills add https://github.com/frabcd/codex-ai-game-studio --skill story-done-frabcd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Closing a development story often skips verification: acceptance criteria go unchecked, GDD or ADR deviations slip in silently, code review is forgotten, and sprint status files drift out of sync. This Skill enforces a structured end-of-story review so a story is only marked Complete after evidence is confirmed. ## Core Features & Use Cases - Acceptance Criteria Verification: Automatically checks file existence, test passes, and hardcoded values, batches manual confirmations, and builds a test-criterion traceability table with COVERED/UNTESTED escalation rules. - Deviation and Evidence Gates: Compares implementation against the TR registry, control manifest version, and ADRs, enforces test evidence requirements per story type (Logic, Integration, Visual/Feel, UI, Config/Data), and runs QA coverage and lead-programmer review gates in full mode. - Status Updates and Sprint Continuity: Updates the story file to Complete, logs tech debt, refreshes sprint-status.yaml and session state, suggests a git commit, and surfaces the next ready story or the sprint close-out sequence. - Use Case: After implementing a damage calculator story in a Godot project, run the review to confirm all criteria pass, the unit test exists, no GDD deviations exist, then close the story and pick up the next ready sprint item. ## Quick Start Use the story-done skill to review and close the story at production/epics/core/story-damage-calculator.md and report the verdict.

Frequently Asked Questions about story-done

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

FAQPage Schema
How do I verify acceptance criteria before marking a story complete?

Run the story-done review on the story file. It automatically checks file existence, test passes, and hardcoded values, batches manual confirmations for subjective criteria, and produces a pass/fail/deferred report with a COMPLETE, COMPLETE WITH NOTES, or BLOCKED verdict.

How to check test coverage for a user story in a sprint?

The review maps each acceptance criterion to a unit test, integration test, or confirmed manual playtest in a traceability table. If more than half the criteria are untested, the verdict is blocked until coverage improves.

Does story completion work without a full QA review process?

Yes. Three review modes control gate behavior: solo skips QA and code review gates, lean prompts you to confirm a manual code review, and full spawns qa-lead and lead-programmer subagent gates before completion.

What happens when implementation deviates from the GDD or ADR?

Deviations are categorized as BLOCKING, ADVISORY, or OUT OF SCOPE and presented neutrally. Blocking deviations prevent a COMPLETE verdict, while advisory ones can be logged to docs/tech-debt-register.md with your approval.

Can a story be marked complete without user approval?

No. The skill never edits files without explicit confirmation. After presenting the completion report, it asks how to proceed: close the story, log tech debt, fix issues first, or accept deviations and close anyway.