faffter-noon-review

Reviews code diffs against specs across five passes and emits a structured review verdict.

3|Updated Apr 16, 2026
One-click install
npx skills add https://github.com/shftwst/faff --skill faffter-noon-review-shftwst
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: faffter-noon-review
Source: https://github.com/shftwst/faff/tree/main/plugin/skills/faffter-noon-review
Command: npx skills add https://github.com/shftwst/faff --skill faffter-noon-review-shftwst

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Automated delivery pipelines need a deterministic pre-PR code review that checks acceptance-criteria coverage, bugs, scope, and spec fidelity before burning CI time, and that routes human-judgement decisions back to a person instead of silently shipping them. ## Core Features & Use Cases - Five review passes: AC coverage, obvious-bug scan, scope check, spec fidelity, and human-judgement flagging, each mapped to a closed verdict vocabulary (pass, fixable-issues, needs-human). - Pre-PR gating: Runs after build and local tests but before the PR is raised, so review-fix iterations never reach CI; findings land on the tracker issue as a single collapsed comment. - Deterministic contract output: Appends a faff-contract:review-verdict JSON block so the orchestrator parses the verdict without re-reading prose. - Use Case: In an autonomous faff-graft run, after a feature branch builds and tests pass, this skill reviews git diff main...HEAD against the committed spec and returns a verdict that decides whether the pipeline iterates, opens the PR, or parks the work for a human. ## Quick Start Let faff-graft invoke this review automatically after a build completes, or configure it explicitly as the review slot in your pipeline's slots YAML.

Frequently Asked Questions about faffter-noon-review

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

FAQPage Schema
How do I automate code review before opening a pull request?▼

Run a pre-PR review pass over the git diff after build and local tests succeed but before the PR is raised. This skill performs five passes (AC coverage, bugs, scope, spec fidelity, human judgement) and returns a verdict the pipeline uses to iterate, open the PR, or park the work.

What does an automated code review check against a spec?▼

It verifies every acceptance criterion has a test or observable verification, scans for common defects like unused imports and unhandled async errors, confirms the diff stays in scope, and checks the implementation matches the spec's stated decisions and error handling.

When should a code review escalate to a human instead of iterating?▼

Escalate when the diff involves decisions the spec did not anticipate, such as new user-facing copy, security posture changes, or irreversible external effects. The test is whether git revert fully undoes the change; if effects persist after revert, it needs a human.

Can I swap the default review skill for a different reviewer?▼

Yes. This skill is the implicit default for the review slot but can be replaced in the slots configuration with a delegated skill such as gstack:review or an alternative review implementation, as long as it conforms to the review-verdict contract.

Why does automated review happen before the PR instead of after?▼

Pre-PR review avoids CI costs from failed runs and idle waiting, since review-fix iterations never reach CI. Findings that would have been PR comments are reported to the tracker issue instead, so no information is lost.