What problem does it solve?
Provides a disciplined, evidence-first workflow to fix bugs, test failures, CI/CD errors, type/lint problems, and UI defects while avoiding symptom-only patches and recurring regressions. It enforces mandatory scouting and structured diagnosis before any fix, and requires verification plus prevention steps so changes are safe, minimal, and auditable.
Core Features & Use Cases
- Structured Workflow: Stepwise scout → diagnose → assess complexity → implement fix → verify & prevent → finalize, with explicit outputs for each step.
- Modes & Routing: Supports autonomous, human-in-the-loop, quick (trivial fixes), deep (research + planning), and parallel multi-agent workflows.
- Tooling & Orchestration: Integrates Task-based orchestration, parallel Explore/Bash agents, and mandatory activation of ck:scout, ck:debug, and ck:sequential-thinking; includes review and docs update gates.
- Prevention Gate: Always captures pre-fix state, requires regression tests or runtime assertions, and applies defense-in-depth where applicable.
- Use Case: Triage a failing test or CI job, trace the root cause across modules, implement a minimal fix with a regression test, then run parallel verification (typecheck, lint, build, tests) before finalizing.
Quick Start
Use the ck:fix skill to run a fast scout→diagnose→fix→verify cycle for a simple type or lint error in the repository.