Review connected MCP servers, which give agents access to external tools, for unused or overlapping capabilities. Get recommendations to keep, disable, or review each server, with estimates of the request size each recommendation could save.
Audit connected MCP servers for token overhead, redundancy, and security. Use when sessions feel slow or before adding new MCPs.
Run score Not tested
See how to installReview the instructions the agent will receive before you install.
Analyze MCP server overhead and recommend cleanup.
Use when:
Each MCP server adds ALL its tool descriptions to every API request. A server with 20 tools adds ~2K-4K tokens per request, regardless of whether you use those tools.
Check all MCP configurations:
cat .claude/settings.json 2>/dev/null | grep -A 50 "mcpServers"
cat ~/.claude/settings.json 2>/dev/null | grep -A 50 "mcpServers"
For each server, estimate token overhead:
Questions to ask:
Disable servers that:
Keep servers that:
MCP AUDIT
Active servers: [N]
Total tools: [N]
Estimated overhead: ~[N]K tokens per request
Server Analysis:
[name] — [N] tools, ~[N] tokens
Status: KEEP / DISABLE / REVIEW
Reason: [why]
Recommendations:
Disable: [list]
Keep: [list]
Review: [list]
Projected savings: ~[N]K tokens per request (~$X.XX per session)
disabledMcpjsonServers to avoid re-recommending already-disabled serversNo reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Coordinate multiple Claude Code sessions on a shared codebase. Organize tasks, dependencies, teammate messages, and file ownership, with guidance on team size, plan approval, and choosing between teams, subagents, and worktrees.
Set up project checks when starting work on an unfamiliar codebase. Detect the project type, verify available tools, and generate Claude Code settings for linting, type checks, tests, formatting checks, and hooks.
Coordinate repetitive changes across many files by splitting them into independent tasks. Agents work in separate Git working directories, run relevant tests, and open pull requests, with a shared summary of successes and failures.
Turn conversational bug reports into GitHub issues with clear reproduction steps, expected behavior, and project terminology. The skill checks for duplicates and separates independent problems, keeping reports focused on behavior so they remain useful after code changes.
Prepare for conversation compaction, which shortens an agent's working context, by saving active files, progress, decisions, blockers, and next steps. Afterward, use those notes and priority files to recover the working state and resume the task.
Manage the information an AI coding session carries by saving notes, selecting relevant material, summarizing results, and separating tasks. The skill guides context budgets and session handoffs so important decisions remain available during long projects.