API Documentation

The LiquidCortex API lets you list AI employees, chat with them, search your knowledge base, and run workflows from your own code.

Quick Start

curl https://liquidcortex.ai/api/employees/ \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json"

curl -X POST https://liquidcortex.ai/api/chat/message \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "employee_id": "atlas",
    "messages": [
      { "role": "user", "content": "Schedule a team standup for tomorrow at 10am" }
    ],
    "stream": false
  }'

The API declares a single security scheme, HTTPBearer. Send your key in the Authorization header of every request.

Request Header

Authorization: Bearer lc_live_sk_xxxxxxxxxxxxxxxxxxxx

API Key Types

Key management endpoints

List all AI employees. Returns a JSON array (no wrapper object, no query parameters).

Response

[
  {
    "id": "atlas",
    "name": "Atlas",
    "role": "Executive Assistant",
    "phase": "...",
    "department": "...",
    "tier": "...",
    "specializations": ["..."],
    "description": "..."
  }
]

Field list is EmployeeInfoResponse from the OpenAPI schema. Values shown as "..." vary per employee; call the endpoint for real data.

Get one employee by ID. Same EmployeeInfoResponse shape as the list.

Path Parameters

Send a message to an AI employee. A session is created automatically when session_id is omitted.

Request Body

Example Request (non-streaming)

{
  "employee_id": "atlas",
  "messages": [
    { "role": "user", "content": "Schedule a meeting with the marketing team for tomorrow at 2pm" }
  ],
  "stream": false
}

Streaming

curl -X POST https://liquidcortex.ai/api/chat/message \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: text/event-stream" \
  -d '{"employee_id": "atlas", "messages": [{"role": "user", "content": "Draft an email..."}], "stream": true}'

The response body is not typed in the OpenAPI document; inspect it in the live Swagger UI.

List conversations for the authenticated user.

Query Parameters

Response

{
  "conversations": [ /* ConversationListItem[] */ ],
  "total": 0,
  "limit": 50,
  "offset": 0,
  "has_more": false
}

Get a conversation with its message history (message_limit default 100, max 500; message_offset for paging).

Update a conversation's title or summary.

Delete a conversation and all its messages.

Upload a document to the knowledge base.

Form Fields

Response

{
  "id": "...",
  "name": "...",
  "file_type": "...",
  "file_size_bytes": 0,
  "status": "...",
  "message": "...",
  "created_at": "..."
}

List documents (query: status, limit default 50 max 100, offset).

Get one document, including chunk and token counts and processing status.

Delete a document from the knowledge base.

Semantic search across your documents.

Request Body

Example Request

{
  "query": "What is our refund policy?",
  "top_k": 5,
  "similarity_threshold": 0.7
}

Response

{
  "query": "What is our refund policy?",
  "results": [
    {
      "chunk_id": "...",
      "document_id": "...",
      "document_name": "...",
      "content": "...",
      "chunk_index": 0,
      "similarity": 0.0,
      "metadata": {}
    }
  ],
  "count": 1,
  "search_time_ms": 0
}

Create a new workflow.

Request Body

Example: two-step review pipeline

{
  "name": "Lead qualification pipeline",
  "trigger_type": "manual",
  "steps": [
    {
      "order": 1,
      "employee_id": "hunter",
      "action": "analyze",
      "input": "Qualify this lead: {{input.lead}}"
    },
    {
      "order": 2,
      "employee_id": "sterling",
      "action": "generate",
      "input": "Draft a proposal based on: {{prev.output}}"
    }
  ]
}

action is the StepAction enum: analyze, generate, review, execute.

List workflows (query: status = draft | active | inactive | archived, trigger_type, limit, offset).

Get one workflow. PUT updates it; DELETE removes it.

Run a workflow now.

Returns a WorkflowRunResponse: id, status (pending | running | completed | failed | cancelled), steps_completed, steps_total, progress_percent, credits_used, results.

Execution history for a workflow (query: status, limit, offset).

List the integrations that can be connected (id, name, category, description).

List your connected integrations and their status.

Start an OAuth connection. Returns { auth_url, message }; redirect the user to auth_url.

Integration IDs

As returned by GET /integrations/available on 2026-09-15 (21 entries).

Disconnect an integration. Returns { success, message }.

Tool catalog in the MCP spec shape. Requires a live API key (lc_live_sk_ / lc_test_sk_).

Streamable HTTP transport: a JSON-RPC 2.0 envelope with method tools/call. Tools: cortex_pre_task_brief and cortex_post_task_ingest.

Unauthenticated liveness probe.