# humanrail-mcp-server

When your AI hits something it can't handle — content moderation, refund decisions, data verification, subjective assessments — HumanRail routes it to a human worker, verifies the result, pays the wo…

## Quick Start

```bash
# Connect this server (installs CLI if needed)
npx -y smithery mcp add humanrail/humanrail-mcp-server

# Browse available tools
npx -y smithery tool list humanrail/humanrail-mcp-server

# Get full schema for a tool
npx -y smithery tool get humanrail/humanrail-mcp-server create_task

# Call a tool
npx -y smithery tool call humanrail/humanrail-mcp-server create_task '{}'
```

## Direct MCP Connection

Endpoint: `https://humanrail-mcp-server--humanrail.run.tools`

**Optional config:**
- `HUMANRAIL_API_KEY` (header) — Your HumanRail API key (get one at humanrail.dev)

## Tools (7)

- `create_task` — Create a task for human review and judgment. Use when the AI agent needs a human to make a subjective decision — conten…
- `get_task` — Get the current status and result of a task. When status is 'verified', the 'output' field contains the worker's verifi…
- `wait_for_task` — Wait for a task to complete by polling until it reaches a terminal state (verified, failed, cancelled, or expired). Use…
- `cancel_task` — Cancel a task that hasn't been completed yet. Only works for non-terminal states (posted, assigned, submitted). Cannot …
- `list_tasks` — List tasks with optional filters. Useful for checking recent activity, finding completed tasks, or monitoring pending w…
- `get_usage` — Get usage statistics and billing summary for your organization. Returns task counts, API call volumes, and average late…
- `health_check` — Check if the HumanRail API is healthy and reachable. Returns status and HTTP code.

```bash
# Get full input/output schema for a tool
npx -y smithery tool get humanrail/humanrail-mcp-server <tool-name>
```