fix

Repairs build, lint, and type-check failures with minimal verified changes.

Updated Feb 12, 2026
One-click install
npx skills add https://github.com/kieranpotts/skills --skill fix-kieranpotts
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fix
Source: https://github.com/kieranpotts/skills/tree/main/skills/fix
Command: npx skills add https://github.com/kieranpotts/skills --skill fix-kieranpotts

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a build, compiler, linter, or type-checker fails, developers often guess at fixes, over-edit the diff, or silence errors with unexplained suppressions. This Skill enforces a disciplined repair workflow: read the tool's output literally, make the minimal change at the established location, and prove the fix worked by re-running the failing check. ## Core Features & Use Cases - Literal error-driven repair: Reads the exact error, rule name, and location reported by build, lint, and type-check tools, preferring automated fixes like --fix flags and codemods before manual edits. - Diagnosis handover support: Acts on a confirmed causal chain and failing regression test from a prior diagnosis, verifying the test goes green with the fix and red again when reverted. - Guardrails against scope creep: Forbids refactors, renames, weakened tests, and unexplained suppressions riding along in the repair diff. - Use Case: After CI reports eslint failures on a pull request, invoke the Skill to run the linter, apply autofixes, manually resolve remaining violations at their reported locations, and confirm every project check passes before committing. ## Quick Start Fix the lint errors reported in the latest CI run and verify all project checks pass afterward.

Frequently Asked Questions about fix

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

FAQPage Schema
How do I fix failing lint errors automatically?▼

Run the linter with its autofix flag first, such as eslint --fix, then re-run to see what remains. Fix each remaining violation manually at the exact rule and location the tool reports, and re-run the check to confirm zero violations.

How do I verify a bug fix actually worked?▼

Re-run the exact command that originally failed and confirm it exits zero. If a regression test was handed over, watch it pass with the change applied, then revert the change and confirm it fails again, proving the test exercises the fix.

When should I use a lint suppression comment?▼

Suppress only when the rule is genuinely wrong for that specific case, and always name the rule plus the reason inline, such as eslint-disable-next-line no-explicit-any with a justification. Unexplained suppressions hide signal from future readers.

What should I do when the cause of a build failure is unknown?▼

Stop rather than guessing repeatedly. A repair workflow only applies when a tool or completed diagnosis has established the cause; otherwise hand the work to a systematic diagnosis process that forms and tests hypotheses first.

Why does fixing one type error cause many more errors?▼

Narrowing a shared type or bumping a dependency can cascade errors across call sites. Work outward from the original error since downstream failures usually resolve once the upstream cause does, and treat large cascades as a maintenance sweep.