fix

Diagnoses a reported bug, implements the root-cause fix, and ships it to a green pull request.

Updated Jun 15, 2026
One-click install
npx skills add https://github.com/nsollazzo/ship-kit --skill fix-nsollazzo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fix
Source: https://github.com/nsollazzo/ship-kit/tree/main/.agents/skills/fix
Command: npx skills add https://github.com/nsollazzo/ship-kit --skill fix-nsollazzo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a reported bug into a merged-quality pull request normally requires juggling diagnosis, branching, testing, committing, and PR management by hand. This Skill chains that entire loop into one command so a bug report goes straight to a green, mergeable PR. ## Core Features & Use Cases - Parallel-agent triage: Fans sub-agents across reproduction, recent changes, root-cause tracing, and blast-radius analysis, then adversarially verifies the leading hypothesis before any code changes. - Root-cause implementation: Creates a fix/<slug> branch, writes a failing test first, and applies the smallest correct fix at the root cause rather than the symptom. - Full ship pipeline handoff: Delegates simplify, verify, code-review, smart-commit, PR creation, and babysitting to the bundled ship skill, stopping at approved + CI green. - Use Case: A user reports "checkout total is wrong when a coupon is applied." Invoke the skill with that description and it triages the cause, branches, fixes it with a regression test, and delivers a green PR — hands-off with --yolo. ## Quick Start Ask the agent to fix the reported issue by describing the bug, for example: fix the login timeout error users see after password reset, optionally adding --yolo for a hands-off run.

Frequently Asked Questions about fix

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

FAQPage Schema
How do I fix a bug and open a pull request in one command?▼

Invoke the fix skill with the issue description, error message, or failing test as the argument. It triages the root cause with parallel sub-agents, creates a fix branch, implements the fix with a regression test, then hands off to the ship skill which reviews, commits, opens, and babysits the PR to green.

How do I run an autonomous bug fix without manual confirmations?▼

Pass the --yolo flag to run hands-off: triage proceeds autonomously, the fix is implemented without a confirmation pause, and the ship checkpoint auto-confirms. The run stops at a green PR; merging still requires the explicit --merge flag.

Does the fix skill work with AI agents other than Claude Code?▼

Yes, the kit follows the open Agent Skills standard with a SKILL.md entry point, so it installs on any compatible agent including Codex, Grok, Hermes, and VS Code Copilot. Harness-specific fields like argument-hint are simply ignored by other agents.

What happens if the root cause of a bug cannot be found?▼

The skill halts at the triage gate and reports the diagnosis-so-far plus what remains unknown, rather than branching and guessing. It enforces an iron law of no fix without a verified root cause backed by evidence.

Can I resume an interrupted bug fix run?▼

Yes, the skill writes a .fix-state.json file after each stage recording the branch, issue, flags, diagnosis, and completed stages. On the next invocation it offers to resume from the next stage, and under --yolo it resumes automatically without asking.