Coding agents
Use Autohand models in other coding agents
Bring Autohand Fantail and Moa to the coding agent you already use. Each guide takes you from a new API key to a working prompt.
Quick answer
How do I use Autohand models in another coding agent?
Create an API key in Autohand Console, store it in AUTOHAND_API_KEY, and add Autohand to the agent as an OpenAI-compatible provider with the base URL https://api.autohand.ai/v1 and the model ID fantail or moa. Agents that do not speak Chat Completions connect through a local LiteLLM proxy.
- Status
- Available for nine coding agents
- Requires
- An Autohand account and an Autohand API key
- Base URL
https://api.autohand.ai/v1- Model IDs
fantailandmoa
Know before you start: The Autohand API serves the OpenAI Chat Completions format. It does not serve the OpenAI Responses API or the Anthropic Messages API.
Before you begin
Every guide starts with the same two steps:
- Sign in to Autohand Console. If you do not have an account, sign up first. The Free plan includes Fantail, and Moa requires the Pro plan or above.
- Open API Keys, create a key for the agent, and store it in the
AUTOHAND_API_KEYenvironment variable.
Confirm that the key works before you configure the agent. The request lists the models your plan can use.
curl https://api.autohand.ai/v1/models \
-H "Authorization: Bearer $AUTOHAND_API_KEY"Autohand API at a glance
| Setting | Value |
|---|---|
| Base URL | https://api.autohand.ai/v1 |
| Request format | OpenAI Chat Completions: POST /v1/chat/completions |
| Model list | GET /v1/models |
| Authentication | Authorization: Bearer $AUTOHAND_API_KEY |
| Model IDs | fantail and moa. The API also accepts autohand/fantail and autohand/moa. |
Choose a model
| Model ID | Use it for | Context | Max output | Plans |
|---|---|---|---|---|
fantail | Fast agent loops, quick fixes, reviews, and subagents | 256K tokens | 16K tokens | Free and above |
moa | Planning, large refactors, and repository-wide changes | 1M tokens | 262,144 tokens | Pro and above |
Use Fantail for the main loop and for subagents, which run many short tasks. Use Moa when the agent plans a large change or needs the whole repository in context.
Supported coding agents
| Agent | Connects through | Subagents on Fantail |
|---|---|---|
| Codex | Local LiteLLM proxy (Responses API) | [agents] default_subagent_model |
| Claude Code | Local LiteLLM proxy (Anthropic Messages API) | CLAUDE_CODE_SUBAGENT_MODEL and subagent model |
| Pi | Direct custom provider | Subagent example extension |
| Hermes Agent | Direct custom endpoint | delegation.model |
| OpenCode | Direct custom provider | agent.<name>.model |
| Kilo Code | Direct OpenAI Compatible provider | See the guide |
| Cline | Direct OpenAI Compatible provider | See the guide |
| Factory Droid | Direct custom model | See the guide |
| Aider | Direct OpenAI-compatible model | See the guide |
Agents that connect directly send Chat Completions requests to Autohand. Codex and Claude Code use other request formats, so their guides add a local LiteLLM proxy that translates each request.
Next steps
Common questions
Coding agents and Autohand FAQ
Do I need an Autohand account?
Yes. You need an account to create the API key that every agent uses.
Which coding agents work with Autohand models?
Codex, Claude Code, Pi, Hermes Agent, OpenCode, Kilo Code, Cline, Factory Droid, and Aider.
Why do Codex and Claude Code need a proxy?
Codex uses the OpenAI Responses API and Claude Code uses the Anthropic Messages API. The Autohand API serves Chat Completions, so a local LiteLLM proxy translates the requests.