Audit coding-agent token spend with CodeBurn
Run CodeBurn locally or as an MCP server so agents can inspect token usage, cost, model mix, project spend, and budget waste across coding tools.
npx skills add agentskillexchange/skills --skill audit-coding-agent-token-spend-with-codeburn
Use CodeBurn when an operator or agent needs to explain and reduce AI coding spend from local session data. The workflow is concrete: run `npx codeburn`, `codeburn overview`, reports, guard checks, or the `codeburn mcp` server; inspect cost and token usage by provider, model, task, project, tool, activity, and MCP overhead; then use the findings to tune model choice, guardrails, delegation, and workflow habits. Invoke this instead of using product billing dashboards when the question is where coding-agent tokens went across Claude Code, Cursor, Codex, Gemini, and related tools on the developer machine. Keep the boundary on local-first spend analysis and MCP/CLI reporting; do not treat this as a general observability platform, billing product, or proxy.
What this skill actually does
Inputs and prerequisites: Node.js 22.13+; local session data from supported AI coding tools; optional Claude Code, Cursor, or another MCP client for codeburn mcp.
Setup notes: Run instantly with npx codeburn or install with npm install -g codeburn . For agent access, add the MCP server with claude mcp add codeburn — npx -y codeburn mcp or configure another MCP client to run npx -y codeburn mcp over stdio.
Source and verification boundary: use https://codeburn.app/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a MCP workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.