refire

What problem does it solve?

When software breaks—failing tests, production bugs, build breaks, or flaky behavior—teams often rush to quick fixes that mask the underlying issue, leading to repeat incidents, longer downtime, and compounded technical debt, especially under pressure like CI blocking deployments.

Core Features & Use Cases

• Structured Root Cause Investigation: A step-by-step process to trace failures to their source, including reading full error details, reproducing the issue, auditing recent changes, checking documented contracts, and comparing against working examples.
• Fix Guardrails: Requires a written hypothesis, a pinned failing test that asserts the broken contract before any fix, minimal single-variable changes, and automatic escalation after three failed fix attempts to avoid guesswork.
• Common Pitfall Avoidance: Explicitly calls out and rejects common rationalizations for skipping process, like "quick fix now, investigate later" or "just try X and see", that lead to recurring bugs. Use case: If your CI pipeline is failing after a recent dependency update, use this skill to systematically trace the root cause instead of randomly tweaking configs to get a green build that breaks again later.

Quick Start

Use the refire skill to investigate the root cause of the failing end-to-end test suite in the current repository before proposing any code changes.