🚀 Release: mcp-audit v0.5.0

[nathanschram]

🎉 mcp-audit v0.5.0 is here!

Find + fix context bloat & high token usage in your MCP tools.

---

🆕 What's New in v0.5.0

Added

• Smell Detection Engine — Automatically detect 5 efficiency anti-patterns (#16)
  • HIGH_VARIANCE: Token counts vary wildly (CV > 50%)
  • TOP_CONSUMER: Single tool dominates (>50% of tokens)
  • HIGH_MCP_SHARE: Heavy MCP reliance (>80% of tokens)
  • CHATTY: Too many calls (>20 calls per tool)
  • LOW_CACHE_HIT: Cache underutilized (<30% ratio)
• Zombie Tool Detection — Identify MCP tools defined but never called (#17)
  • Configure known tools in mcp-audit.toml
  • Session logs report unused tools per server
• Data Quality System — Accuracy labels for all metrics (#18)
  • exact: Native platform tokens (Claude Code)
  • estimated: Tokenizer-based estimation (Codex CLI, Gemini CLI)
  • calls-only: Only call counts, no tokens (future Ollama CLI)
  • Confidence scores and estimation method metadata
• AI Prompt Export — Export session data for AI analysis (#19)
  • mcp-audit export ai-prompt command
  • Markdown and JSON output formats
  • Includes suggested analysis questions
• Schema v1.5.0 — New data blocks (#21)
  • smells: Detected efficiency anti-patterns with evidence
  • zombie_tools: Unused tools per MCP server
  • data_quality: Accuracy metadata (level, source, confidence)

➡️ View Milestone

➡️ View all v0.5.0 issues

---

⚡ Install or Upgrade

# New install
pipx install mcp-audit

# Upgrade existing
pipx upgrade mcp-audit
# or: pip install --upgrade mcp-audit

---

🔗 Resources

📦 PyPI	pypi.org/project/mcp-audit/0.5.0/
📋 Changelog	Full version history
📚 Docs	README & platform guides
🗺️ Roadmap	What's coming next

---

🤖 Supported Platforms

Platform	Token Accuracy
Claude Code	Native (100%)
Codex CLI	Estimated (99%+)
Gemini CLI	Estimated (100% with tokenizer)

---

💬 Questions, feedback, or want support for another CLI platform?
Reply here or start a discussion!

Made with 🐻 by Little Bear Apps