Tuning Coding Agent Codebases

Audits and refactors a codebase's coding agent configuration (CLAUDE.md, .claude/settings.json, skills, hooks) against Anthropic's published best practices. Primary use is rewriting an existing CLAUDE.md so the root holds only pointers and gotchas under 200 lines, conventions move into subdirectory CLAUDE.md files, procedures migrate from CLAUDE.md into skills, and rules added to compensate for older models get deleted. Also audits .claude/settings.json permission denies, scoped test and lint commands, skill bloat, hook obsolescence against the Claude Code changelog, and the decision matrix for CLAUDE.md vs skills vs hooks vs plugins vs MCP. Works on any codebase size, not just monorepos. Use this skill when the user asks to audit, clean up, refactor, rewrite, or improve CLAUDE.md or Claude Code config, says the agent feels slow or scattered in this repo, runs a periodic config review, sets up a coding agent in a new repo, or asks 'is my setup right'. Trigger even when the user says only 'clean up my CLAUDE.md' or 'review my Claude Code config'.

mkdir -p ~/.claude/skills/tuning-coding-agent-codebases
curl -L https://claudskills.com/skills/tuning-coding-agent-codebases/SKILL.md \
  -o ~/.claude/skills/tuning-coding-agent-codebases/SKILL.md

@misc{tuning-coding-agent-codebases-2026,
  author    = {drag88},
  title     = {Tuning Coding Agent Codebases [Claude Code skill]},
  year      = {2026},
  publisher = {ClaudSkills},
  url       = {https://claudskills.com/skills/tuning-coding-agent-codebases/}
}