🤖 ai kodi

9 Skill Categories for Claude Code (from Anthropic's Internal Use)

From Thariq’s thread on how Anthropic internally categorizes Claude Code skills into 9 types — plus practical tips for writing effective ones.

The 9 Skill Categories

#CategoryExamples
1Library/Tool Referencebilling-lib, internal-cli, frontend-design
2Verificationsignup-flow-driver, checkout-verifier
3Data & Monitoringfunnel-query, grafana, cohort-compare
4Workflow Automationstandup-post, create-ticket, weekly-recap
5Scaffoldingnew-workflow, new-migration, create-app
6Code Qualityadversarial-review, code-style, testing-practices
7Deploymentbabysit-pr, deploy-service, cherry-pick-prod
8Debugging/Incidentoncall-runner, log-correlator, service-debugging
9Maintenance/Opsresource-orphans, dependency-management, cost-investigation

Key Tips for Writing Skills

  1. Skills are folders, not just markdown — Include scripts, assets, reference code, templates
  2. Gotchas section = highest signal — Built from real Claude failure points
  3. Progressive disclosure — Split into references/, assets/, etc. Tell Claude what’s there
  4. Be flexible, not rigid — Give info but let Claude adapt
  5. Description = trigger condition — Not a summary, but “when to use this”
  6. Config.json for user setup — Store per-user settings like Slack channels
  7. Skills can have memory — Log files, JSON, SQLite. Use ${CLAUDE_PLUGIN_DATA} for stable storage
  8. Give Claude scripts — Code > prose for repetitive tasks
  9. Skill-scoped hooks — e.g. /careful blocks dangerous commands only when activated

The takeaway: Skills aren’t just documentation — they’re mini-applications. The best ones combine markdown guidance with executable scripts, persistent memory, and smart trigger conditions. Think of them as giving Claude a specialized toolkit, not just instructions.