mattpocock/skills: An Agent skill set for real-world engineering

The project splits skills into two types: user-triggered skills invoked manually via /commands (such as /grill-with-docs, /triage, /improve-codebase-architecture, /to-spec, /implement), and model-triggered skills the Agent calls automatically when a task matches (such as tdd, diagnosing-bugs, research, domain-modeling, code-review). The core idea constrains four common failure scenarios: misaligned requirements via repeated questioning with /grill-me; inconsistent terminology via a shared language doc to cut verbosity; unreliable code via red-green refactoring and /diagnosing-bugs to build a feedback loop; and architecture rot via /improve-codebase-architecture to periodically scan for deep-module opportunities. Installation differs by tool: Claude Code uses claude plugin install with auto-updates; Codex and GitHub Copilot add plugins via the marketplace; Gemini CLI uses gemini skills install with manual updates; other Agents or cases needing editable files use npx skills add with manual updates. After installing, run /setup-matt-pocock-skills once per repo to configure the issue tracker, triage labels, and doc location. For developers writing code daily with Claude Code, Codex, Copilot, or Gemini CLI, this turns engineering discipline—requirement alignment, test-driven development, architecture review—into reusable commands, rather than relying on ad-hoc prompts in every new session.