What problem does it solve?
Multiple engineering skills for Claude Code (including triage, to-issues, and tdd-hybrid) require consistent repo-specific context such as the issue tracker type, triage label vocabulary, and domain documentation layout. Without centralized configuration, each skill either guesses, asks repetitive onboarding questions, or assumes incorrect defaults (like GitHub for repos that use GitLab), leading to inconsistent behavior and wasted user effort.
Core Features & Use Cases
- Auto-detection of existing repo context: Inspects git remotes, existing AGENTS.md/CLAUDE.md files, and current docs structure to avoid asking redundant questions.
- Sequential guided configuration: Walks users through three key decisions (issue tracker, triage labels, domain docs layout) one at a time with plain-language explainers of why each choice matters for downstream skills.
- Centralized config output: Writes a shared
## Agent skills block to AGENTS.md/CLAUDE.md and three reference files under docs/agents/ that all downstream engineering skills read automatically.
- Real-world use case: A team using GitLab with custom triage labels (like
bug:triage instead of the default needs-triage) can map their actual label vocabulary once, and the triage skill will apply the correct labels without creating duplicates.
Quick Start
Use the setup-matt-pocock-skills skill to configure your repo's shared context for all downstream engineering skills before first use of tools like triage, to-issues, or tdd-hybrid.