Reliant MCP

Control your Reliant AI reliability layer by voice or chat — directly from Claude Desktop, Claude Code, or any MCP-compatible client.

What it does

Once installed, you can talk to Claude and say things like:

• "Check the health of my Reliant project"
• "Create a guard for my WhatsApp bot that can only talk about our products, and limits each user to 50 messages per day"
• "Run this invoice text through my extraction schema"
• "What's the SLA status of my invoice-extraction schema?"
• "Show me the last 20 executions that failed"
• "Create a business contract that ensures the extracted amount is always positive"

Installation

1. Install the package

npm install -g reliant-mcp

2. Configure Claude Desktop

Open your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the Reliant MCP server:

{
  "mcpServers": {
    "reliant": {
      "command": "reliant-mcp",
      "env": {
        "RELIANT_API_KEY": "rel_your_api_key_here",
        "RELIANT_API_URL": "https://reliant-production.up.railway.app",
        "RELIANT_USER_ID": "your-supabase-user-id"
      }
    }
  }
}

3. Get your credentials

• RELIANT_API_KEY — Dashboard → Settings → API Key
• RELIANT_API_URL — Dashboard → Settings → API URL
• RELIANT_USER_ID — Dashboard → Settings → User ID

4. Restart Claude Desktop

The Reliant tools will appear automatically.

Available tools

| Tool | Description | |------|-------------| | reliant_health_check | Quick health check of your entire setup | | reliant_metrics | Summary metrics — executions, success rate, latency | | reliant_analytics | Analytics by schema and provider | | reliant_execute | Run a prompt through a schema | | reliant_list_executions | List recent executions | | reliant_inspect_execution | Inspect a specific execution in detail | | reliant_list_schemas | List all schemas | | reliant_create_schema | Create a new schema by describing its fields | | reliant_get_schema | Get schema definition and config | | reliant_guard_chat | Send a message through a Guard | | reliant_list_guards | List all Guard configs | | reliant_create_guard | Create a Guard by describing its rules | | reliant_guard_usage | Check Guard usage stats | | reliant_audit_recent | Recent events from the Audit Trail | | reliant_inspect_execution | Inspect a specific execution | | reliant_sla_status | Check SLA status of a schema | | reliant_sla_configure | Set up SLA targets for a schema | | reliant_sla_list | List all SLAs | | reliant_create_contract | Create a Business Contract |

Example conversations

Health check

"How is my Reliant setup doing?"

Claude will call reliant_health_check and show API status, schema count, today's success rate, and any SLA breaches.

Create a guard

"Create a guard for my customer support bot. It should only talk about our products, never mention competitors, and always be professional. Limit each user to 100 messages per day."

Claude will call reliant_create_guard with the appropriate config and return the guard ID.

Check SLA

"Check the SLA for my invoice extraction schema"

Claude will call reliant_sla_status and show current success rate, latency, and whether targets are being met.

Run an extraction

"Extract the invoice data from this text: 'Invoice #4521, amount $12,500, client: Acme Corp, due date: 2026-06-15' using my invoice-extraction schema"

Claude will call reliant_execute and return the structured JSON output.

Use with Claude Code

Add to your project's .mcp.json:

{
  "mcpServers": {
    "reliant": {
      "command": "reliant-mcp",
      "env": {
        "RELIANT_API_KEY": "rel_...",
        "RELIANT_USER_ID": "..."
      }
    }
  }
}

License

MIT — Celeiro Softwares · reliant.ia.br