fix

Implement safe code fixes using a guardrail-driven three-pass workflow.

66|2|Updated Feb 18, 2015
One-click install
npx skills add https://github.com/tkersey/dotfiles --skill fix
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fix
Source: https://github.com/tkersey/dotfiles/tree/main/codex/skills/fix
Command: npx skills add https://github.com/tkersey/dotfiles --skill fix

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Fix skill provides a structured protocol to review and implement code changes safely, ensuring guardrails for unsoundness, invariants, footguns, and incidental complexity, and requiring a validation signal before delivering changes.

Core Features & Use Cases

  • Guarded triage: Triage issues by severity (security, crash, logic) and outline the smallest valid fix.
  • Multi-pass workflow: Execute a three-pass safety, surface, and audit process with pass traces and proof hooks.
  • Proof-oriented validation: Capture invariant_before and invariant_after, counterexamples, and proof results for each finding.
  • Embedded governance: Support embedded-mode fix records and strict deliverable formats for PRs.

Quick Start

Use the fix skill to create a safe, validated patch for a failing repository state and verify it with a local signal.

Frequently Asked Questions about fix

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

FAQPage Schema
How do I implement safe code fixes with validation guardrails?

A guardrail-driven protocol enforces validation signals before completing code fixes. You run a deterministic three-pass workflow—Safety, Surface, Audit—prioritizing correctness, compatibility, then performance, while generating pass traces and proof hooks for each finding.

What is a guardrail-driven protocol for validating code changes?

A guardrail-driven protocol validates code changes by capturing invariant_before and invariant_after states, collecting counterexamples, and requiring proof results. It triages issues by severity like security, crash, or logic, and outlines the smallest valid fix before execution.

How do I triage code issues by severity before applying a patch?

Triage code issues by categorizing severity into security, crash, and logic buckets. You then outline the smallest valid fix for each finding, capturing invariant states and counterexamples to ensure the patch maintains correctness before execution.

Can I use invariant tracking and counterexamples to verify code reproductions?

Invariant tracking verifies code reproductions by defining invariant_before and invariant_after states. You collect counterexamples for each finding and generate proof results, ensuring the patch maintains correctness across footguns and incidental complexity.

What's the best way to audit code changes for incidental complexity and footguns?

Auditing code changes for incidental complexity and footguns is best handled through a three-pass workflow. The Audit pass follows Safety and Surface passes, evaluating performance while producing pass traces and proof hooks for each identified finding.

Do I need a local validation signal to complete a guarded code fix?

A local validation signal is required to complete a guarded code fix. The protocol enforces this validation signal before delivering any changes, ensuring the patch passes the Safety, Surface, and Audit workflow successfully.