ryoppippi/ccusage v17.0.0

on GitHub

What's Changed

✨ New Features

• Full OpenAI Codex support via new @ccusage/codex package
• Added 1M token context window pricing support for Sonnet

⚠️ Breaking Changes

• npx ccusage mcp ➡️ npx @ccusage/mcp

🚀 Performance

• 2/3 bundle size reduction
• Migrated from Zod to Valibot for improved performance

📝 Installation

# Codex Analysis
bunx @ccusage/codex@latest  # ⚠️ Must include @latest with bunx
npx @ccusage/codex           # npx/pnpx work without @latest

# MCP Server  
npx @ccusage/mcp@latest --type http --port 8080 # for remote mcp
npx @ccusage/mcp@latest # for stdio

# The Original
bunx ccusage
npx ccusage

⚠️ Important for bunx users: When using bunx with @ccusage/codex, you MUST include the @latest suffix (e.g., bunx @ccusage/codex@latest).

Reason: Bun 1.2.x's bunx has a behavior where it prioritizes binaries matching the package name suffix when given a scoped package. For example, with @ccusage/codex, bunx looks for a codex binary in PATH first. If you have an existing codex command installed, that will be executed instead of the intended package. Adding @latest forces bunx to fetch and run the correct package. This issue does not affect npx or pnpx.

   🚀 Features

• Monorepo(bun -> pnpm)  -  by @ryoppippi in #631 (4d8b2)
• Extract MCP server into dedicated package  -  by @ryoppippi in #632 (6a594)
• Codex  -  by @ryoppippi in #641 (fead5)
• ccusage: Migrate from Zod to Valibot for schema validation  -  by @ryoppippi in #633 (bdd88)
• codex: Add session command for Codex token usage analysis  -  by @ryoppippi in #645 (93646)
• docs: Add ESLint configuration for documentation linting  -  by @ryoppippi in #644 (37aec)
• internal: Add support for 1M context window pricing  -  by @ryoppippi in #651 (e8e46)
• mcp: Add package badges to README  -  by @ryoppippi in #637 (5795c)

   🐞 Bug Fixes

• Update example ccusage config  -  by @ryoppippi (6a678)
• Bin for development  -  by @ryoppippi in #636 (40286)
• ccusage/statusline: Cache is negatable  -  by @ryoppippi (6c90e)
• internal: Fix context size bug by using max_input_tokens instead of max_tokens  -  by @ryoppippi in #650 (6e617)

    View changes on GitHub