Harness
Everything wrapped around the model that turns "it can talk" into "it can work".
What is a harness?
An AI model on its own can only talk: you ask, it answers, it stops. The harness is all the machinery wrapped around the model that lets it actually do work: tools, memory, safety rails, and the loop that keeps it going until the job is done. Claude Code, Cursor, and your agent runtimes are all harnesses around the same few models. ClawMetry is a window into that machinery. Each part below links to the tab where you can watch it live. (Good primer: What's harness engineering?)
🔁 The loop
The agent reads, decides, acts, checks its own work, and goes again until the task is done. This loop is what makes it an agent instead of a chatbot.
Watch it live in Activity →
🔧 Tools
Tools let the agent touch the world: read and edit files, run code, search the web, call other services. No tools, no work.
See its tools →
📚 Memory
Models forget everything between sessions. The harness writes notes to files so your agent remembers you, your projects, and its own lessons.
Read its memory →
🪟 Context
The model can only "see" a limited window of text at once. The harness keeps that window focused, trimming old noise so quality doesn't rot.
See context usage →
📦 Sandbox
A safe room where the agent can run code and fail freely, without reaching things it shouldn't touch.
Check its sandbox →
🚦 Guardrails
The rules for what the agent may do on its own and what needs your OK first, like deleting files or spending money.
Review approvals →
🤝 Teamwork
For big jobs, one agent leads and hands pieces to helper agents, then stitches the results together.
See who spawned whom →
💬 Interfaces
The same agent can meet you in a terminal, on Telegram or Slack, or in a browser. Different doors, one harness behind them.
Follow the message flow →
What this runtime uniquely exposes, beyond the generic tabs:
Loading harness view…