Bug Fix — Stop-the-Line Protocol

Trust Score 86/100

Structured bug-fix workflow and triage protocol: reproduce, localize, reduce, fix, add regression test, and verify before resuming development.

triggers:fix bugdebugfailing teststop the lineregression

GitHub SKILL.md

What it does

This skill codifies a conservative, test-first bug-fix workflow intended to reduce regressions and speed reliable fixes. It defines a Stop-the-Line protocol (preserve evidence, diagnose causality, fix root causes only, and add regression tests) plus a Triage Sequence to reproduce and narrow failures before implementing changes. The guidance is practical for developer-facing agents assisting with debugging and PRs.

When to use it

Use when a test, build, lint, or runtime error appears and a safe, auditable fix is required. Ideal for CI failures, flaky tests, pre-merge checks, and situations where regressions are costly. This skill should be triggered by phrases like "fix bug", "debug", "resolve issue", or when an agent detects failing tests or CI alerts.

What's included

Scripts: none (has_scripts=false) — the skill references repo contribution guides and testing patterns but provides procedural instructions rather than executable scripts.
References: inline links to contribution docs and testing-patterns in the repo.
Instructions: step-by-step protocol (Stop-the-Line) and a Triage Sequence, plus a quality checklist and red flags to avoid.

Compatible agents

Developer-assistant agents that can read/write repo files, run tests, and create PRs (GitHub-enabled Copilot/gh-cli agents, Claude/Codex-style agents).

Audit Summary

Structured bug-fix workflow skill with a stop-the-line protocol, triage sequence, and anti-rationalization guidance. No bundled scripts — purely instructional SKILL.md. Well-written with practical debugging traps and a quality checklist. Assumes pnpm and gh CLI tooling, which limits portability but covers the common JS/TS project case.

Watch Out

Assumes pnpm as package manager — not portable to npm/yarn projects
gh pr create hardcodes --base main, may need adjustment for different default branches
allowed-tools in frontmatter is project-specific and may not match all agent runtimes

Notes

Solid instructional skill. The anti-rationalization table is a standout feature — genuinely useful for agents and humans alike. Architecture could improve by extracting the workflow checklist into references/ and adding a script for automated triage. The 'Co-Authored-By: Claude' in commit template is agent-specific but not a security concern.

Information

Repository: nexus-agents
Stars: 10
Installs: 0

Trust Score

Overall86

Security95

Code Quality82

Architecture68

Usefulness72

Related Skills

Development Worktree

Create an isolated git worktree for feature work, auto-run project setup, and verify a clean test baseline before development.

ds-fix — data-science mid-analysis fixer

Orchestrates diagnosis and targeted fixes mid-analysis: diagnose root cause, apply fixes with output-first verification, and update project learnings.

Full Stack Builder

End-to-end builder that scaffolds, implements, tests, and optionally deploys web and API applications from a natural-language specification.

Claw Bench

Benchmarking skill that guides an agent through a structured suite of capability tests and reporting steps for leaderboard submission.

Tmux Bridge

Give an agent controlled access to a local tmux terminal bridge so it can run shell commands, capture output, and manage sessions on the user's machine.

Replit — Advanced Troubleshooting

Step-by-step diagnostics and debugging techniques for deep Replit issues: Nix build failures, container crash loops, memory leaks, and platform vs app isolation

Juicebox Common Errors

Identify and remediate common Juicebox API errors (authentication, quotas, rate limits, invalid queries) with quick diagnostics and fixes.

Vercel Common Errors

Diagnose and fix common Vercel build, serverless runtime, and edge/routing errors with step-by-step checks and fixes.

Back to Skills