ClawMetry
vivekchand/clawmetry
Local dashboard that reads agent session files to show timelines, tool calls and token costs
ClawMetry is an observability dashboard that reads the session files your coding agents already write and shows timelines, tool calls, and token and cost data in one view. Hermes is one of the more than 30 agent runtimes it supports.
What ClawMetry does
ClawMetry runs locally and opens at http://localhost:8900. It reads session files and logs read-only, with no SDK, code change or instrumentation in your app, and shows session timelines, tool-by-tool replay, token and cost breakdowns, and trajectory signals such as looping and repeated failures, per runtime. When several runtimes are active, a header switcher re-scopes every tab to one of them.
Hermes appears in the README's list of runtimes that need the closed-source clawmetry-pro companion, which arrives with a 7-day trial or a plan. The free pip package reads OpenClaw, NVIDIA NemoClaw, Goose and Qwen Code with no account, key or network call. The README says no session data leaves your machine unless you run clawmetry connect, and that an anonymous install ping and a PyPI version check run by default and can be switched off. It also notes that runtimes expose very different data, and some publish no cost at all.
Key features
- Session transcripts and turn-by-turn replay
- Token and cost breakdown per runtime and model
- Trajectory signals such as looping and repeated failures
- A header switcher that scopes every tab to one runtime
- A sample mode with three labelled synthetic sessions
- Read-only access to session files with no code changes
When to use it
- Telling a long agent run that is working from one that is stuck
- Comparing token spend across several agent runtimes on one machine
- Replaying what an agent did tool call by tool call
Who it is for: Developers who run one or more coding agents and want a single view of sessions, tool calls and costs.
How it fits with Hermes Agent
It supports Hermes Agent among more than 30 runtimes, reading its sessions through the paid clawmetry-pro companion rather than the free open-source app.
How to install ClawMetry
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
pip install clawmetry && clawmetryNote: Reading Hermes sessions depends on the closed-source clawmetry-pro companion, which arrives with a 7-day trial or a paid plan.
FAQ
What is ClawMetry?
ClawMetry is an observability and governance dashboard for AI agent runtimes. It reads the session files agents already write and shows timelines, tool calls, tokens and costs.
Does ClawMetry work with Hermes Agent?
Yes. Hermes is on the list of supported runtimes, but the README places it among those read by the paid clawmetry-pro companion rather than the free open-source app.
How do I install ClawMetry?
Run pip install clawmetry && clawmetry, then open http://localhost:8900. The README says clawmetry --sample opens three synthetic sessions if you have no agents installed yet.
Similar security for Hermes Agent
All securitySecurity assistant framework for authorized bug bounty work, with an MCP server and Hermes Agent support
EXboys EvotownSelf-hosted control plane for governing OpenClaw, Hermes and SkillLite agents across a company
runta-dev ClawShellLocal proxy that swaps virtual API keys for real ones and scans traffic for PII, for OpenClaw and Hermes
0xNyk LACPLocal policy, approval and recovery controls that wrap CLI coding agents such as Claude, Codex and Hermes
asimons81 Hermes VaultLocal-first credential broker, secret scanner and encrypted vault for Hermes agents
78 TenBoxLightweight virtual machine monitor for running OpenClaw, QwenPaw and Hermes Agent in isolation
Related guides: How to run Hermes Agent securely