fix

Diagnose and repair failing tests, type errors, lint failures, and review findings.

4|1|Updated Jul 31, 2025
One-click install
npx skills add https://github.com/alvis/.agents --skill fix-alvis
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fix
Source: https://github.com/alvis/.agents/tree/main/plugins/coding/skills/fix
Command: npx skills add https://github.com/alvis/.agents --skill fix-alvis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reproducible failures like broken tests, type errors, lint violations, and CI failures need disciplined repair without introducing new behavior or patch seams; this Skill diagnoses the root cause and restores green checks while respecting the active plan contract. ## Core Features & Use Cases - Failure Diagnosis: Auto-detects the failure area by running tests, type checker, and linter, mapping each error to its code location. - Test and Fixture Repair: Fixes incorrect test logic, mocks, and fixtures without modifying the source code under test. - Plan-Contract Pinning: Pins the state.md plan identity after a code review so follow-up reviews validate against the identical task contract. - Use Case: After a code review flags defects in a stacked change, run the fix workflow to repair the owning change, re-run all checks, and return a structured attempt report with root-cause classification. ## Quick Start Fix the failing tests and type errors in the current work item and report the root cause of each failure.

Frequently Asked Questions about fix

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

FAQPage Schema
How do I fix failing tests without changing source code?

Diagnose the failure by running the test suite, then repair incorrect test logic, imports, fixtures, and mocks only. The workflow restricts edits to test files and test support files, never modifying the source code under test.

When should I use a fix workflow versus refactoring?

Use the fix workflow when a concrete failure can be reproduced, such as failing tests, type errors, lint failures, or review findings. Route structural cleanup of working code to refactoring and new functionality to a code-writing workflow.

Can the fix skill handle review findings from a code review?

Yes, it pins the identical plan contract from state.md so the follow-up review validates against the same task identity. It rejects any plan identity that does not match the recorded plan source.

What happens if the plan definition changes during a fix?

The workflow stops and returns the affected task IDs with a needs_revalidation outcome instead of applying a stale fix. It never mutates the plan while fixing and requires the plan definition to remain unchanged before requesting done status.

Why does the fix workflow reject requests when all checks pass?

The skill requires a reproducible failure or identified defect as input. When all checks pass and no defect exists, there is nothing to repair, so the request is rejected to avoid unnecessary changes.