claude-automation-recommender

Invoke this Skill and it scans your codebase for languages, frameworks, databases, and tooling, then outputs the top 1-2 recommendations per category: MCP servers, skills, hooks, subagents, and plugins. Each recommendation includes the reason and setup location.

Hooks handle automatic actions on tool events like formatting on save, subagents run specialized parallel reviews, skills package repeatable workflows, plugins bundle related skills, and MCP servers integrate external tools like databases and GitHub. The Skill matches each type to signals detected in your code.

No, it is strictly read-only. It analyzes the codebase and outputs a recommendations report, but does not create or modify any files. You implement the recommendations yourself or ask Claude separately to help build them.

Yes. If you ask for a specific type, the Skill focuses only on that category and provides 3-5 options instead of the default 1-2 per category. You can also ask for more recommendations in any category afterward.

The Skill checks package.json, pyproject.toml, and similar manifests for frameworks and libraries, looks for testing and CI/CD configs, detects databases like Supabase or PostgreSQL, and inspects existing .claude/ configuration. Each signal maps to specific MCP servers, hooks, skills, or subagents.