TokenTelemetry

TokenTelemetry provides a local-first observability dashboard designed specifically for AI coding agents. It consolidates data from the log files generated by these agents, offering a unified interface to surface information such as token usage, large language model (LLM) costs, tool calls, session traces, and reasoning steps. The platform is intended for users who want to monitor and analyze the activities and resource consumption of AI coding agents without relying on external cloud services.

The dashboard works entirely on the user's machine, ensuring that logs, prompts, and cost information remain private and never leave the local environment. TokenTelemetry reads existing log files without modifying them, and all agent data is stored in a dedicated directory on the user's system. There are two optional outbound features—an hourly update check and another unspecified feature—both of which are content-free and can be disabled. The tool supports a range of AI coding agents, including Claude Code, Codex, Gemini CLI, Cursor, Copilot, Qwen, OpenCode, Vibe, Antigravity, Grok Build, Cline, SmallCode, Pi, and Hermes Agent.

TokenTelemetry is free to use and open-source, requiring no signup or cloud account. It is launched with a single command, emphasizing ease of installation and use. This approach is aimed at developers and others working with AI coding agents who require insight into agent behavior and resource consumption while maintaining full control over their data and privacy.