TokenTelemetry — Track token cost & reasoning traces across your autoresearch runs (100% local, open source)

[VasiHemanth]

Hey autoresearch community!

Running long agentic research loops is exciting — but it can be a black box: how many tokens did that run burn? What did the agent actually reason through? Which sub-agents fired?

I built TokenTelemetry to answer exactly that — a 100% local, open-source observability dashboard for AI coding agents.

What it tracks

• Token usage & cost per session and per model, in real-time
• Full reasoning traces — see the chain of thought behind every decision
• Subagent activity — which agents spawned, what they did, what they cost
• Tool calls & artifacts — every action taken during the run

Why it's useful for autoresearch

Autoresearch runs multi-step, iterative research loops that can get expensive fast. TokenTelemetry gives you per-run observability — so you can:

• Compare cost across different research strategies
• See exactly where tokens are being spent (planning vs. searching vs. writing)
• Understand reasoning quality vs. cost tradeoffs

Supported agents (zero config)

Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, Qwen, Vibe, Antigravity, OpenCode — reads directly from local log files, no proxies.

Install

curl -fsSL https://raw.githubusercontent.com/VasiHemanth/tokentelemetry/main/install.sh | bash

MIT licensed · 100% local · no signup · Node 18+, Python 3.9+

GitHub: https://github.com/VasiHemanth/tokentelemetry
Website: https://tokentelemetry.com

Would love feedback from folks running deep research loops — especially around multi-agent cost visibility!