Coding Agents
Use Stream with a Coding Agent
Point Hermes, OpenClaw, Claude Code, Codex, OpenCode, or Pi at camelStream's frontier intelligence
Create a key in the Stream console, then run the installer. It detects the agents on your machine, writes the configuration below, and sends one request to confirm it works.
curl -fsSL https://stream.camelai.com/install | bashThe guides on this page are the manual equivalent. No camelStream plugin is
required either way. Every guide uses the same model ID, auto; the model
serving your stream comes from the fleet. See the
installer page for what it changes and how to undo it.
| Model ID | auto |
| OpenAI base URL | https://stream.camelai.com/v1 |
| Anthropic base URL | https://stream.camelai.com |
| Key | CAMEL_API_KEY |
| Context window | 260K tokens guaranteed (see the fleet) |
hermes model, then pick Custom endpoint.
Add a models.providers entry and restart the gateway.
Set the Anthropic gateway and map model aliases to auto.
Add a named Responses profile in ~/.codex.
/connect, choose Other, then select the model.
Add a provider in ~/.pi/agent/models.json.
These are third-party agents. camelStream provides the model. Keep the
key in an environment variable or the agent's secret file, not in a git
repo. Requests beyond your concurrent stream allowance
queue instead of failing. Set up before the
auto model ID existed? Your existing config keeps working; see the
fleet page FAQ for how the legacy ID is handled.