API & MCP

Automate the platform programmatically using a Bearer API key. The primary interface is an MCP server with 46 tools covering agents, runs, access requests, tools, credentials, models, templates, and destinations — but most tools are administrative. Regular customers can run agents, view their own agents and runs, and manage their own credentials; admins have the full set.

Authentication

All API endpoints require a Bearer token in the Authorization header:

Authorization: Bearer YOUR_API_KEY

Create and manage API keys from API Keys in the dashboard sidebar (https://app.tirelesscrew.com/api-keys). Each key belongs to the user who creates it, and every MCP action is attributed to that owner — admin-only tools require an admin's key. Keys are shown only once at creation (only a SHA-256 hash is stored), so store them securely.

Access by role

Every MCP action is attributed to the API key's owner, and the tools available are limited by that owner's role. Calling a tool your role is not permitted to use returns an error.

Role	What you can do over MCP
Customer	Run agents and read your own runs (`run_agent`, `get_run_status`, `list_agent_runs`), view agents you have access to (`list_agents`, `get_agent`), manage your own credentials (`create_credential`, `update_credential`, `delete_credential`, `upsert_credential`, `list_credentials`, `list_credential_templates`), and request access to published agents (`request_agent_access`).
Support	Everything a customer can read, plus running agents — read-only browsing of agents, tools, templates, destinations and credentials. No create, update, or delete.
Admin	The full set of tools — creating, editing, deleting and publishing agents; attaching tools; and managing tools, models, templates, destinations, credentials, and access requests.

The reference tables below list every tool. Tools marked Admin only — creating/editing/deleting/publishing agents, attaching or detaching tools, managing tools, models, destinations, and platform credentials — require an administrator's API key.

MCP server

The platform exposes a Model Context Protocol (MCP) server at:

POST https://app.tirelesscrew.com/api/mcp

This endpoint follows the MCP spec 2025-03-26 streamable HTTP transport: JSON-RPC requests return a response, notifications (no id) get a silent 202, and initialize returns an Mcp-Session-Id header. You can connect it to Claude Desktop / Claude Code (via mcp-remote), or any MCP-compatible client. See the in-app Admin → Docs → MCP Server Setup for a ready-to-paste Claude Desktop config.

Calling a tool

curl -X POST 'https://app.tirelesscrew.com/api/mcp' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -d '{
    "jsonrpc": "2.0",
    "id":      1,
    "method":  "tools/call",
    "params": {
      "name":      "list_agents",
      "arguments": {}
    }
  }'

MCP tools

46 tools are available, grouped below by class. The authoritative list is always what tools/list returns.

Agents

Tool	Description
`list_agents`	List all agents you have access to; includes published_at.
`get_agent`	Full agent detail including tools, destinations summary [{id,type,is_active}], and published_at.
`create_agent`	Create a new agent with a name, model, system prompt, and trigger type.
`update_agent`	Update an existing agent's name, model, system prompt, or trigger.
`delete_agent`	Soft-delete an agent. It is removed from listings but can be restored by an admin.
`attach_tool`	Attach a tool to an agent with an alias and optional credential source.
`detach_tool`	Remove a tool from an agent by its alias.
`publish_agent`	Publish an agent to the catalog; enforces AgentPreflightService, returns checklist on hard failure. Admin only.
`unpublish_agent`	Remove an agent from the catalog. Admin only.
`preflight_agent`	Run preflight checks and return the full checklist without publishing. Admin only.

Agent Runs

Tool	Description
`run_agent`	Trigger an agent run immediately. Performs preflight and credit checks. Returns a run_id to poll.
`get_run_status`	Poll a run for status, output, tokens, cost, duration, and human-readable error.
`list_agent_runs`	List runs scoped to the authenticated user. Filter by agent_id or status. Admins see all runs.

Access Requests

Tool	Description
`request_agent_access`	Submit an access request for a published agent. Customer role only. Rejects duplicate pending requests.
`list_access_requests`	List access requests by status (default: pending). Admin only.
`approve_access_request`	Approve a pending request; grants access and sends a notification email. Admin only.
`deny_access_request`	Deny a pending request and send a notification email. Admin only.

Tool Library

Tool	Description
`list_tools`	Browse all active tools available on the platform.
`get_tool`	Retrieve full configuration for a single tool by slug or ID.
`create_tool`	Create a new tool definition; accepts credential_id and credential_source_preference. Admin only.
`update_tool`	Update an existing tool's configuration including auth fields. Admin only.

Credentials

Tool	Description
`list_credentials`	List your saved credentials.
`create_credential`	Create a new named credential with encrypted field values.
`update_credential`	Update an existing credential's name, label, or config values.
`delete_credential`	Delete a credential. Cannot be undone.
`upsert_credential`	Create or update a credential by name (idempotent). Useful for automation.
`list_credential_templates`	List auth template schemas — the field definitions for each auth type.
`list_admin_credentials`	List platform-scoped credentials. Admin only.
`create_admin_credential`	Create a platform-scoped credential shared across all agents. Admin only.

Reasoning Models

Tool	Description
`list_reasoning_models`	List all configured AI models available on the platform.
`get_reasoning_model`	Retrieve full connection config for a single model.
`create_reasoning_model`	Create a new model with endpoint, auth, and request format config. Admin only.
`update_reasoning_model`	Update an existing model's configuration. Admin only.

Agent Templates

Tool	Description
`list_agent_templates`	List all active agent templates.
`deploy_agent_template`	Deploy a template to create a new agent in your account from a template slug.

Destinations

Tool	Description
`list_destinations`	List all output delivery destinations attached to an agent.
`create_destination`	Add an email, webhook, or Slack destination to an agent.
`delete_destination`	Remove a destination from an agent.

REST endpoints

Two lightweight REST endpoints are also available for model discovery and direct chat completion:

Endpoint	Description
`POST https://app.tirelesscrew.com/api/v1/chat`	Send a chat completion request through the gateway. Pass `model`, `messages`, and optional `options`. Returns generated content and token usage.
`GET https://app.tirelesscrew.com/api/v1/models`	Returns all available model identifiers with per-million-token pricing. Public — no auth required.

Error codes

Status	Meaning
`401`	Missing or invalid `Authorization: Bearer` header.
`402`	Insufficient credits to complete the request.
`403`	The API key does not have permission for this operation (e.g. non-admin attempting an admin-only tool).
`404`	The requested resource does not exist or you do not have access to it.
`422`	Validation error — the request body is missing required fields or contains invalid values.
`429`	Rate limit exceeded. The API allows 60 requests per minute per key.
`500`	Unexpected server error. If this persists, contact support.