> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orcapods.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Use Orca from coding agents

> Run orca mcp serve so Claude Code, Cursor, and Codex can manage and talk to your Orca agents.

`orca mcp serve` runs an MCP server on stdio with one tool per CLI action. A coding agent can then list and create agents, start sessions and chat with them, manage skills, vaults, files, and kits, and read usage and your wallet, all as you.

It uses the CLI's own login, read again on every call, so you can add it before you sign in. Sign in first with [`orca login`](/cli/install#sign-in). Without a key, each tool answers "not logged in to Orca." and how to log in.

## Claude Code

Install the Orca plugin. It adds the `use-orca` skill and registers the MCP server for you:

```bash theme={"dark"}
claude plugin marketplace add okikorg/orca-cli
claude plugin install orca@orca
```

Or register only the MCP server:

```bash theme={"dark"}
claude mcp add orca -- orca mcp serve
```

## Cursor

Add the server to `~/.cursor/mcp.json` for every project, or `.cursor/mcp.json` for one:

```json theme={"dark"}
{
  "mcpServers": {
    "orca": {
      "command": "orca",
      "args": ["mcp", "serve"]
    }
  }
}
```

## Codex

```bash theme={"dark"}
codex mcp add orca -- orca mcp serve
```

Or add it to `~/.codex/config.toml`:

```toml theme={"dark"}
[mcp_servers.orca]
command = "orca"
args = ["mcp", "serve"]
```

## The tools

| Area | Tools |
| - | - |
| Identity and keys | `whoami`, `list_keys`, `create_key`, `revoke_key` |
| Agents | `list_agents`, `get_agent`, `create_agent`, `update_agent`, `delete_agent` |
| Sessions | `list_sessions`, `get_session`, `create_session`, `list_session_items`, `delete_session`, `chat` |
| Skills | `list_skills`, `get_skill`, `create_skill`, `delete_skill` |
| Vaults | `list_vaults`, `create_vault`, `delete_vault`, `list_credentials`, `add_credential`, `delete_credential` |
| Files | `list_files`, `upload_file`, `download_file`, `delete_file` |
| Usage and billing | `get_usage`, `list_usage_events`, `get_wallet`, `billing_checkout`, `billing_portal` |
| Kits | `list_kits`, `make_kit`, `edit_kit`, `publish_kit`, `withdraw_kit`, `show_kit`, `copy_kit` |

`chat` sends a message to an agent or session and waits for the reply, by default up to 120 seconds and at most 600. It returns the session ID, whether the turn finished, its status, and the reply, so the coding agent can continue the same session.

Commands that only touch your machine (login, logout, context, doctor, update) have no tool. Tools act with your key's role, so admin-only actions fail for a member.

## In practice

Ask your coding agent things like:

* "Create an Orca agent called reviewer on `orca/openrouter/openai/gpt-5.6-luna` that reviews pull request descriptions."
* "Chat with reviewer about this diff and give me its answer."
* "How much Orca credit do I have left?"

Treat what the agent does with these tools as you would any command it runs: it can create, delete, and spend on your behalf.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.