Skip to main content
An agent is a first-class, reusable definition of what runs — a name, a prompt (its instructions), a model, and a runtime. Every session runs on an agent, so create one first, then start as many sessions on it as you like. 
curl https://api.opencomputer.dev/v3/agents \
  -H "Authorization: Bearer $OPENCOMPUTER_API_KEY" \
  -d '{ "name": "reviewer", "model": "anthropic/claude-opus-4-8",
        "prompt": "You review code.", "runtime": "claude", "key": "sk-ant-…" }'
Pass your Anthropic key inline and it’s stored as a credential and attached to the agent — sessions run the model on it. Already have a credential? Reference it with credential: "cred_…" instead, or omit both to use your org default. Optionally set default limits (tokens/turn_seconds/turns) too; sessions inherit them.

​
Pinned snapshot

When a session is created it freezes the agent’s prompt, model, runtime, and revision (a number that bumps on every edit) into the session — and pins the resolved credential too. Editing the named agent afterwards never changes a running or resumed session — its behavior is pinned for its whole life. Update an agent freely: existing sessions keep what they started with, new sessions pick up the change. (Rotating a credential’s key value is the exception — that flows to running sessions, so a compromised key can be replaced; see authentication.)

​
Model

model is a provider/model id (e.g. anthropic/claude-opus-4-8); the chosen runtime determines which models are valid. The model runs on your Anthropic key (required). Today the only callable runtime is claude.
Additional model providers, and switching a session’s model mid-session, are coming soon.