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

# merv-client CLI

> The onboarding CLI that configures a machine's connection to the brain and its local agent platforms.

Installed by the runner installer at `~/.merv/bin/merv-client` (also at `merv/bin/merv-client` in a checkout). It configures the connection and the optional local agent platforms used by [Auto-run](https://rapidreview.io/docs/merv/guides/auto-run.md). Interactive sign-in is not here — that belongs to each client's native MCP OAuth command.

```bash theme={"theme":{"light":"github-light","dark":"dark-plus"}}
CLI=$HOME/.merv/bin/merv-client
```

| Command | What it does |
| - | - |
| `$CLI configure [--control-url <url>]` | Write machine config — e.g. which brain to target. Settings live in the private `~/.merv/client.json`. |
| `$CLI env` | Print the header-based `.mcp.json` HTTP snippet for a headless client on this machine. |
| `$CLI agent <platform> --enable\|--disable [--adapter <a>] [--command <exe>] [--model <m>] [--effort <e>] [--parallelism <n>]` | Enable or disable a local agent platform. Commands are stored as argv arrays and never run through a shell. |
| `$CLI agents` | Print the configured local platforms. |
| `$CLI workspace --repository <path> --strategy git_worktree [--root <dir>] [--base-ref <ref>]` | Set the repository, worktree strategy, worktree root, and base ref the runner uses. |
| `$CLI harness [--json]` | Install the Merv skills under `~/.merv/skills/` and check each enabled platform: executable resolves, version, how it reaches Merv tools and skills. Non-zero exit if an enabled platform isn't ready. |
| `$CLI storage-upload --path <file> --target-url <url> [--workers 4]` | Multipart upload helper used by the returned storage upload commands. |
| `$CLI call <TOOL> --arguments <JSON>` | Call a Merv MCP tool from a child agent without a shell — used by the runner bridge for agents without native MCP isolation. |

## `merv-agent-runner`

| Command | What it does |
| - | - |
| `merv-agent-runner` | Dispatch for the paired project read from `client.json`. |
| `merv-agent-runner --project <id>` | Override the project — the headless-key path (`MERV_MCP_KEY`) and loopback brains without auth. |
| `merv-agent-runner --once` · `--poll-seconds <n>` · `--config <path>` | Run one dispatch cycle; change the poll interval; use another `client.json`. |
| `merv-agent-runner pair` | Start a fresh pairing on an already-installed machine. |

## Other executables

| Executable | Purpose |
| - | - |
| `merv-runs-wait` | The long-run watcher — [Long runs](https://rapidreview.io/docs/merv/clients/long-runs.md) |
| `merv-http` · `merv-control` | The brain, local and hosted — [Deploy the brain](https://rapidreview.io/docs/merv/self-host/deploy-the-brain.md) |

## Removed

`merv-client login`, `link`, `links`, `route`, and `unlink` are gone. Sign in through your client's native OAuth command; headless project reach comes from a static grant plus explicit `project_id` arguments.


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