OpenAI builds and deploys advanced AI models like GPT-4o for autonomous agents and workflows.
Token Telemetry
FreeStarting price
0Popularity
About Token Telemetry
Token Telemetry is a local, read-only observability tool for AI coding agents and autonomous systems. It analyzes the log files that agents such as Claude Code, Codex, Cursor, and Hermes already generate without requiring code changes or SDK integration. The tool installs via a single command, runs offline on the user’s machine, and serves a dashboard on localhost:3000 that displays token usage, costs, reasoning traces, and project artifacts. It supports 17 agents and detects them automatically with zero configuration, providing unified visibility into multiple autonomous systems through a single interface. The application is fully open-source under the MIT license and does not upload any user data, logs, or prompts to external servers. Anonymous usage statistics are optional and can be disabled. Token Telemetry is designed for developers who want to monitor, debug, or optimize AI agent performance without cloud dependencies or vendor lock-in.
Key features
- Reads existing agent logs without modification
- Displays token usage, costs, and reasoning traces
- Supports 17 agents including Claude Code, Codex, Cursor, and Hermes
- Local dashboard served on localhost:3000
- Anonymous usage statistics (optional)
- Open-source with MIT license
- Detects agents automatically with zero configuration
- Read-only access to logs and data
Use cases
- Monitoring token consumption and costs of AI coding agents
- Debugging reasoning traces and subagent delegation in autonomous systems
- Comparing performance across multiple AI tools in a single interface
Pros
- Fully local and offline operation
- No SDK or code instrumentation required
- Supports 17 AI coding agents and autonomous tools
- Open-source under MIT license
- No signup or API keys needed
Cons
- Requires Node 18+ and Python 3.9+
- Limited to desktop environments
- No cloud or remote access to dashboard
Frequently asked questions about Token Telemetry
What is Token Telemetry and what does it do?
Token Telemetry is a local, read-only observability tool for AI coding agents and autonomous systems. It analyzes log files generated by agents such as Claude Code, Codex, Cursor, and Hermes without requiring code changes or SDK integration, providing visibility into token usage, costs, reasoning traces, and project artifacts.
Who is Token Telemetry designed for?
The tool is designed for developers who want to monitor, debug, or optimize the performance of AI agents without relying on cloud dependencies or vendor lock-in. It is particularly useful for those using multiple autonomous systems and seeking unified visibility.
How does Token Telemetry work?
Token Telemetry reads the log files that AI agents already generate on the user’s machine, processes them locally, and serves a dashboard on localhost:3000. It requires no instrumentation, code changes, or SDK integration, and operates entirely offline.
Does Token Telemetry require any setup or configuration?
No configuration is needed. Token Telemetry automatically detects supported agents and scans their log files with zero setup. Installation is done via a single command, and the tool runs locally without requiring an account or API keys.
What agents does Token Telemetry support?
Token Telemetry supports 17 agents, including Claude Code, Codex, Cursor, Gemini CLI, Antigravity, Qwen CLI, Vibe, Copilot, OpenCode, Grok Build, Cline, SmallCode, Pi Muse Code, Prime Agent, DeepSeek Harness, Hermes Agent, and Nous Research.
Does Token Telemetry send any user data to external servers?
No. Token Telemetry operates entirely offline and does not upload logs, prompts, or user data to external servers. Only optional, anonymous usage statistics (e.g., feature usage) may be sent, which can be disabled in the settings.
