skills

Spawns an Explore subagent to map architecture, layers, dependencies, and tech debt hotspots across the entire codebase, then generates a visual HTML report with a Mermaid architecture diagram and opens it in the browser. Finishes with a targeted grilling loop on the findings. Make sure to use this skill whenever the user says "analyze my repo", "what's wrong with my codebase", "architecture review", "technical debt audit", "how is this codebase structured", runs /analyze-repo, or asks for any codebase-level assessment — even phrased casually like "take a look at how this is organized".

Guides system design and architecture decisions: scaling strategies, monolith vs microservices, REST vs GraphQL, SQL vs NoSQL, component boundaries, data flow, rendering strategies, design patterns, code organization, and ADRs. Use proactively whenever the user asks "how should I structure this", "what's the right approach for", mentions tech stack decisions, component design, state management, or code organization — even if they don't say the word "architecture".

Audits and improves CLAUDE.md files to make Claude more compliant and effective. Use when the user says "optimize my CLAUDE.md", "my claude.md is too long", "improve my claude instructions", "claude isn't following my instructions", or any time they want to improve how Claude behaves in their project. CLAUDE.md is loaded on every turn — bloated instructions dilute the signal.

Creates semantic git commits with conventional commit format, stages changes, and pushes to remote. Handles pre-commit hooks and writes meaningful commit messages. Make sure to use this skill whenever the user says "commit", "push changes", "save to git", "commit this", or wants to create a git commit — even if they just say "save my work." Also proactively offer after any implementation task wraps up.

Debugs errors, test failures, and unexpected behavior with systematic root-cause analysis. Use when the user reports a bug, says "this isn't working", a test is failing, an exception is thrown, output is wrong, or asks to investigate "why does X happen". Applies to frontend, backend, database, network, and performance issues.