minhtran3124/agent-harness/intent-review
- community
Checking that the SKILL.md file is still in the repository.
Independently review a finished diff against the user's verbatim original request. Use after correctness review to detect missing requested behavior, intent drift, and unapproved excess scope.
Install
npx vskill@latest install minhtran3124/agent-harness/intent-reviewSKILL.md
Intent Review
This is the intent oracle: it is neither a runtime-bug hunt nor a style review. Run it standalone with a supplied intent, or after correctness review in SDD before finishing a branch.
Establish a blind oracle
- Read
### Intentfromspecs/<slug>/SUMMARY.md; it is the verbatim primary oracle. - Optionally read design success criteria and PLAN §3 SC rows only as secondary checks. A missing
SUMMARY intent and design intent is a STOP: ask the user instead of reconstructing intent.
- If a secondary oracle conflicts with verbatim intent, verbatim wins and the conflict is drift.
- Determine
BASE..HEAD(task-start to HEAD in SDD, merge-base to HEAD for a branch, or an
explicit range). Give the reviewer the oracle, diff, touched files, SC table, and SUMMARY Verify table—but never PLAN.md prose or research-brief.md.
When SDD supplies REVIEW_PACKAGE_PATH, the reviewer reads that mechanical BASE..HEAD package first instead of rebuilding the same diff. The package carries no intent oracle; do not use it to read PLAN prose or research material. Standalone calls without a package retain explicit-range construction.
Dispatch intent-reviewer-prompt.md in fresh context. Every finding must quote the intent sentence it evaluates.
Route findings
- gap: a requested outcome—or a promised SC with no mapped passing Verify row—is missing.
Fix if unambiguous and in scope; otherwise escalate.
- drift: implementation differs from requested behavior. Record equivalent drift as advisory;
route material behavior changes like gaps.
- excess: unrequested behavior or public surface. Report only: do not remove shipped behavior
without human approval.
Before any fix routing, read `rules/auto-correct-scope.md`. Put ambiguous gaps/drift and intent conflicts in ESCALATIONS.md; put advisory drift, excess, and deferred findings in SUMMARY.md under ### Intent Findings. A finding must be fixed with its commit SHA or durably recorded before handoff; otherwise completion is blocked.
The plan/spec reviewer is plan-oriented and correctness-review is runtime-oriented. Keep this oracle separate and plan-blind; it is the check for “implemented correctly, but not what was asked.”
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/intent-review/SKILL.md
Popularity
- Repository stars
- 7
- 7-day trend score
- 10
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/minhtran3124/agent-harness/intent-review)