Documentation / API Reference

API Reference

Interact with Parabable programmatically via the MCP server.

MCP Server Endpoint

The Parabable MCP server uses SSE (Server-Sent Events) as its transport protocol. All tool invocations go through a single endpoint:

POST /mcp/sse

The server accepts JSON-RPC 2.0 requests and streams responses back via SSE. For simple request-response patterns, the frontend's API proxy at /api/tools/* provides a standard JSON REST interface.

Authentication

Authenticate requests using a Bearer token in the Authorization header:

Authorization: Bearer <your-token>

Tokens are provisioned per-client. The frontend uses an internal token configured at deploy time. For programmatic access, request a token from the platform administrator.

Available Tools

The MCP server exposes 18 tools across five categories. Each tool accepts named arguments and returns structured JSON results.

Graph Tools

query_cypher	Execute a raw Cypher query against the graph database
get_node	Retrieve a single node by ID with all properties
get_neighbors	Get adjacent nodes and relationships for a given node
search_nodes	Full-text search across node labels and properties

Memory Tools

store_memory	Store a text memory with embedding for semantic retrieval
search_memory	Semantic search over stored memories
get_memory	Retrieve a specific memory by ID
delete_memory	Remove a memory from the store

Analysis Tools

get_communities	Detect community clusters in the social graph
get_influence_scores	Calculate PageRank-style influence for agents
get_sentiment_timeline	Aggregate sentiment over ticks
get_action_breakdown	Count actions by type for a simulation run

Authoring Tools

create_scenario	Define a new simulation scenario
create_agent	Add an agent to an existing scenario
run_simulation	Start a simulation run for a scenario
seed_content	Inject initial posts or products into a scenario

System Tools

health	Check server and backend connectivity
list_tools	Enumerate all available MCP tools

Example Tool Call

Tool calls use the JSON-RPC 2.0 format with tools/call as the method:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "query_cypher",
    "arguments": {
      "query": "MATCH (a:Agent) RETURN a.name, a.archetype LIMIT 10"
    }
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "content": [
      {
        "type": "text",
        "text": "[{\"a.name\": \"Agent_001\", \"a.archetype\": \"Influencer\"}, ...]"
      }
    ]
  }
}

Health Endpoint

A lightweight health check is available at:

GET /api/health

Returns the server status, version, tool count, and backend connectivity for graph, memory, and embedding services. Returns HTTP 200 when healthy.

Metrics Endpoint

Prometheus-compatible metrics are exposed at:

GET /api/metrics

Key metrics include request latency, tool call counts, active simulations, graph node count, and memory store size. These feed into the Observability dashboard.

← Concepts Tutorials →