TokenTelemetry — See exactly what your accomplish sessions cost, reason & do (100% local observability, open source)

[VasiHemanth]

Hey accomplish community!

Accomplish is all about letting AI agents handle complex coding tasks autonomously — but when those sessions run, do you know what they're actually spending and reasoning through?

I built TokenTelemetry — a 100% local, open-source observability dashboard that gives you full visibility into your AI coding agent sessions.

What it tracks

• Token usage & cost per session, per model, in real-time
• Reasoning traces — the full chain of thought your agent used to solve the task
• Subagent activity — which sub-agents spun up, what they did, what they cost
• Tool calls & artifacts — every action taken during the session
• Real-time dashboard — monitor active sessions live

Why it pairs perfectly with accomplish

When accomplish runs an agentic task, TokenTelemetry shows you the complete picture underneath — not just the final result, but how the agent got there, how much it cost per step, and what reasoning led to each decision. Great for:

• Debugging sessions that went off-track
• Optimizing prompts to reduce cost
• Understanding agent behavior patterns

Supported agents (zero config)

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

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

Happy to hear feedback from accomplish users — would love to know what observability features would matter most to you!

[richardchen874-sys]

TokenTelemetry's local session-level cost trace is the concrete cost boundary here; it shows whether the AI feature is becoming more useful or just more expensive.

I would separate raw usage collection from cost presentation: first normalize tokens and provider metadata per request, then let dashboards group by user, workflow, or model tier.

I am testing a multi-model OpenAI-compatible API layer around official Chinese models, so the routing and usage-accounting parts of this are directly relevant to me. For coworker, do you need billing-grade accuracy here, or is operational usage visibility enough for the first version?