Self Integration

1a. Check for existing connections

GET /connections

Look for a connection matching the target app. Key fields: id, name, connectorId, disconnected.

If a matching connection exists and disconnected is false, skip to Step 2.

1b. Find a connector

A connector is a pre-built adapter for an external app. Search by app name:

GET /search?q=slack

Look for results with elementType: "connector". Use element.id as connectorId in step 1d.

If nothing is found, go to step 1c to build a connector.

1c. Build a connector (if none exists)

Create a Membrane Agent session to build a connector:

POST /agent/sessions with body {"prompt": "Build a connector for Slack (https://slack.com)"}

Adjust the prompt to describe the actual app you need. Poll GET /agent/sessions/{sessionId}?wait=true&timeout=30 until state is "idle" or status is "completed".

You can send follow-up instructions via POST /agent/sessions/{sessionId}/message or abort via POST /agent/sessions/{sessionId}/interrupt.

After the connector is built, search for it again (step 1b).

1d. Request a connection

Create a connection request so the user can authenticate with the external app:

POST /connection-requests with body {"connectorId": "cnt_abc123"}

The response includes a url. Tell the user to open the url to complete authentication (OAuth, API key, etc.).

1e. Check connection result

Poll until the user completes authentication:

GET /connection-requests/{requestId}

• status: "pending" — user hasn't completed yet, poll again.
• status: "success" — done. Use resultConnectionId as the connection ID going forward.
• status: "error" — failed. Check resultError for details.

Step 2: Get an Action

An action is an operation you can perform on a connected app (e.g. "Create task", "Send message", "List contacts").

2a. Search for actions

Search using a natural language description of what you want to do:

GET /actions?connectionId=con_abc123&intent=send+a+message&limit=10

Each result includes id, name, description, inputSchema (what parameters the action accepts), and outputSchema (what it returns).

If no suitable action exists, go to step 2b.

2b. Build an action (if none exists)

Use Membrane Agent. ALWAYS include the connection ID in the prompt:

POST /agent/sessions with body {"prompt": "Create a tool to send a message in a channel for connection con_abc123"}

Adjust the prompt to describe the actual action you need. Poll for completion the same way as step 1c. After the action is built, search for it again (step 2a).

Step 3: Run an Action

Execute the action using the action ID from step 2 and the connection ID from step 1:

POST /actions/{actionId}/run?connectionId=con_abc123 with body {"input": {"channel": "#general", "text": "Hello!"}}

Provide input matching the action's inputSchema.

The result is in the output field of the response.

API Reference

Base URL: ${MEMBRANE_API_URL:-https://api.getmembrane.com} Auth header: Authorization: Bearer $MEMBRANE_TOKEN

GET /connections

List all connections.

Response:

{
  "items": [
    {
      "id": "string",
      "name": "string",
      "connectorId": "string",
      "integrationId": "string (optional)",
      "disconnected": "boolean",
      "state": "NOT_CONFIGURED | SETUP_IN_PROGRESS | SETUP_FAILED | READY",
      "error": "object (optional)",
      "createdAt": "datetime",
      "updatedAt": "datetime"
    }
  ]
}

GET /search

Search workspace elements by keyword.

Query parameters:

Response:

{
  "items": [
    {
      "elementType": "Connector",
      "element": {
        "id": "string",
        "name": "string",
        "logoUri": "string (optional)"
      }
    }
  ]
}

POST /connection-requests

Create a connection request for user authentication.

Request body (at least one identifier required):

Response:

{
  "requestId": "string",
  "url": "string",
  "status": "pending | success | cancelled | error",
  "connectorId": "string (optional)",
  "integrationId": "string (optional)",
  "resultConnectionId": "string (optional, set on success)",
  "resultError": "object (optional, set on error)",
  "createdAt": "datetime"
}

GET /connection-requests/:requestId

Check connection request status. Same response schema as POST.

GET /actions

List or search actions.

Query parameters:

Response:

{
  "items": [
    {
      "id": "string",
      "name": "string",
      "key": "string",
      "description": "string (optional)",
      "type": "string",
      "inputSchema": "JSON Schema (optional)",
      "outputSchema": "JSON Schema (optional)",
      "integrationId": "string (optional)",
      "connectionId": "string (optional)"
    }
  ]
}

POST /actions/:actionId/run

Run an action.

Query parameters:

Request body:

Response:

{
  "output": "any"
}

POST /agent/sessions

Create an agent session to build connectors or actions.

Request body:

Response:

{
  "id": "string",
  "status": "queued | starting | running | completed | failed | cancelled",
  "state": "busy | idle",
  "prompt": "string",
  "createdAt": "datetime",
  "updatedAt": "datetime"
}

GET /agent/sessions/:id

Get agent session status.

Query parameters:

Response: same schema as POST /agent/sessions.

POST /agent/sessions/:id/message

Send a follow-up message to an active agent session.

Request body:

Response: same schema as POST /agent/sessions.

POST /agent/sessions/:id/interrupt

Abort an agent session.

Response:

{
  "interrupted": "boolean"
}

External Endpoints

All requests go to the Membrane API. No other external services are contacted directly by this skill.

Security & Privacy

• All data is sent to the Membrane API over HTTPS.
• MEMBRANE_TOKEN is a high-privilege credential that can create connections and run actions across external apps. Treat it as a secret.
• Connection authentication (OAuth, API keys) is handled by Membrane — credentials for external apps are stored by the Membrane service, not locally.
• Action inputs and outputs pass through the Membrane API to the connected external app.

By using this skill, data is sent to Membrane. Only install if you trust Membrane with access to your connected apps.