Tenant MCP server
The Tenant MCP server exposes your TrainAR account as a set of structured tools that AI models and agents can call directly -- the same operations available in…
The Tenant MCP server exposes your TrainAR account as a set of structured tools that AI models and agents can call directly -- the same operations available in the REST API, but wrapped in a protocol designed for language model tool-use.
What MCP is
MCP (Model Context Protocol) is an open standard, developed by Anthropic, that lets AI assistants call external tools in a consistent, typed way. Instead of writing custom API glue code for each AI platform, you register an MCP server once and any compatible client -- Claude Desktop, Claude Code, n8n AI nodes, OpenAI-compatible agents, or your own code -- can discover and call its tools automatically.
Connection details
| URL | mcp.trainar.ai/sse |
| Transport | Server-Sent Events (SSE) |
| Auth | Authorization: Bearer tak_your_key_here header |
Your tak_* API key scopes apply to MCP calls exactly as they do to REST calls. A key with only read:tasks cannot create a task via the MCP server. Create keys with appropriate scopes in Dashboard → API & Webhooks → API Keys.
Connect from Claude Desktop
Add the following to your claude_desktop_config.json (usually at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"trainar": {
"url": "https://mcp.trainar.ai/sse",
"headers": {
"Authorization": "Bearer tak_your_key_here"
}
}
}
}
Restart Claude Desktop. The TrainAR tools will appear in the tools panel when you start a new conversation.
Connect from Claude Code
Add the server to your project or global Claude Code config:
claude mcp add trainar \
--url https://mcp.trainar.ai/sse \
--header "Authorization: Bearer tak_your_key_here"
Or add it manually to .claude/settings.json:
{
"mcpServers": {
"trainar": {
"url": "https://mcp.trainar.ai/sse",
"headers": {
"Authorization": "Bearer tak_your_key_here"
}
}
}
}
Connect from n8n
Use the MCP Client node (available from n8n 1.x with the AI Agent node) or the HTTP Request node pointed at the SSE endpoint. For most TrainAR use cases in n8n, the TrainAR community node is simpler -- it pre-wires the most common trigger and action operations without needing MCP configuration.
If you want an AI agent in n8n to have full TrainAR tool access, add the MCP server URL and API key to your AI Agent node's tools configuration.
Connect from custom agents or ChatGPT
Any MCP-compatible client can connect. Pass the API key as a Bearer token in the Authorization header when establishing the SSE connection. For clients that use the JSON-RPC over HTTP POST style, the same header applies.
For OpenAI-based agents, use a tool-calling wrapper that forwards MCP tool schemas to the OpenAI function-calling interface and routes calls back to the SSE endpoint.
Tool catalogue
The Tenant MCP server exposes ~70 tools across the categories below. Call tools/list from any MCP client to get the live schemas with full parameter descriptions — that's the source of truth, this page is a high-level overview.
Api Keys (3 tools)
| Tool | Description |
|---|---|
list_api_keys |
List all API keys for your tenant |
create_api_key |
Create a new API key with specified scopes |
revoke_api_key |
Revoke an existing API key |
Billing (7 tools)
| Tool | Description |
|---|---|
get_subscription |
Get current subscription details |
create_subscription |
Start a new subscription |
update_subscription |
Change subscription plan or billing details |
cancel_subscription |
Cancel the active subscription |
get_billing_details |
Retrieve billing contact and payment information |
get_usage_balance |
Check remaining minutes in the current period |
get_minute_transactions |
List minute usage transactions |
Devices (5 tools)
| Tool | Description |
|---|---|
list_devices |
List all paired AR devices in your tenant |
get_device |
Get details for a specific device |
generate_pairing_code |
Generate a code to pair a new device |
unpair_device |
Remove a device from the tenant |
list_orders |
List hardware orders associated with the tenant |
Integrations (6 tools)
| Tool | Description |
|---|---|
list_integrations |
List available integration providers |
list_tenant_integrations |
List integrations connected to your tenant |
get_integration_connect_url |
Get the URL to begin a connection flow |
get_integration_status |
Check the connection status of a specific integration |
disconnect_integration |
Disconnect an integration |
trigger_sync |
Manually trigger a sync for a connected integration |
Knowledge (14 tools)
| Tool | Description |
|---|---|
list_knowledge |
List knowledge items in your tenant |
get_knowledge |
Retrieve a specific knowledge item |
upload_knowledge |
Upload a new knowledge item |
upload_knowledge_file |
Upload a file to a knowledge item |
bulk_upload_knowledge_files |
Upload multiple files at once |
process_knowledge |
Trigger processing/indexing of a knowledge item |
search_knowledge |
Semantic search across your knowledge base |
update_knowledge_file |
Update metadata on a knowledge file |
delete_knowledge |
Delete a knowledge item |
list_knowledge_collections |
List knowledge collections |
get_knowledge_collection |
Get a specific knowledge collection |
create_knowledge_collection |
Create a new collection |
update_knowledge_collection |
Update a collection's metadata |
delete_knowledge_collection |
Delete a collection |
Reports (5 tools)
| Tool | Description |
|---|---|
get_reports_overview |
High-level summary of tenant activity (mirrors the Reports → Overview tab) |
get_sessions_report |
Session activity breakdown for a period |
get_users_report |
Per-user session activity |
get_minutes_report |
Pool balance + usage-over-time |
get_tasks_report |
Task creation, completion, status breakdown |
Sessions (3 tools)
| Tool | Description |
|---|---|
list_sessions |
List training sessions with optional filters |
get_session |
Get full details for a specific session |
get_session_analytics |
Aggregated analytics for sessions |
Skills (6 tools)
| Tool | Description |
|---|---|
list_skills |
List all skills available to your tenant |
get_skill |
Get details for a specific skill |
create_skill |
Create a new skill |
update_skill |
Update an existing skill |
delete_skill |
Delete a skill |
execute_skill |
Execute a skill with arguments |
Tasks (5 tools)
| Tool | Description |
|---|---|
list_tasks |
List tasks with optional status/source/assignee filters |
get_task |
Get a specific task by ID |
create_task |
Create a new task |
update_task |
Update task fields or status |
delete_task |
Delete a task |
Account (2 tools)
| Tool | Description |
|---|---|
get_account |
Get current tenant account info (name, locale, plan summary) |
update_account |
Update tenant-level fields like name and timezone |
get_account_settings |
Retrieve tenant configuration |
Users (6 tools)
| Tool | Description |
|---|---|
list_users |
List users in your tenant |
get_user |
Get details for a specific user |
invite_user |
Invite a new user by email |
update_user_role |
Change a user's role |
deactivate_user |
Deactivate a user account |
reactivate_user |
Reactivate a previously deactivated account |
Webhooks (4 tools)
| Tool | Description |
|---|---|
list_webhooks |
List registered webhook endpoints |
create_webhook |
Register a new webhook endpoint |
update_webhook |
Update an endpoint's URL or events |
delete_webhook |
Remove a webhook endpoint |
Tool schemas
This page lists tools by name and purpose. For the full parameter schemas -- field names, types, and which are required -- call tools/list from your connected MCP client. Claude Desktop and Claude Code display these automatically in the tools panel; other clients can request them via the standard MCP tools/list method.
Scope enforcement
The MCP server enforces the same scope rules as the REST API. If an agent attempts to call invite_user with a key that lacks the manage:users scope, the tool call returns a permission error. Design your key's scopes to match what the agent actually needs.
Next steps
- API keys and permission scopes -- configure the right scopes for your agent.
- Tenant REST API reference -- for direct HTTP calls without an MCP client.
- Webhooks -- complement the MCP server with event-driven delivery for real-time reactions.