Quickstart

The API is public, read-only, and requires no key. Search the content catalog with one request:

curl "https://prescientai.com/api/v1/search?q=halo%20effects&limit=5"

Every successful list response includes a page.next_cursor. Pass that value as cursor to continue. Records have stable IDs for direct retrieval.

REST endpoints

GET /api/v1API capabilities and canonical links.

GET /api/v1/searchSearch all public content; accepts q, type, cursor, and limit.

GET /api/v1/integrationsList or filter supported integrations.

GET /api/v1/content/{id}Retrieve one record using a URL-encoded stable ID.

POST /api/v1/batchRun up to 25 read operations in one idempotent request.

POST /api/v1/jobsCreate a read-only async search job; poll the returned status URL.

GET /api/v1/healthCheck service availability and version.

The complete parameter and response schemas are published at /openapi.json and mirrored at /api/openapi.json.

Model Context Protocol servers

Prescient publishes two stateless MCP servers using Streamable HTTP and JSON-RPC 2.0. Both support current MCP 2026-07-28 discovery plus legacy 2025 initialization, tool discovery, tool calls, resources, structured errors, and behavioral annotations.

https://prescientai.com/mcpProduct integrations, pricing, evaluation planning, and hypothetical budget scenarios.

https://prescientai.com/mcp/docsDocumentation search and content retrieval.

curl https://prescientai.com/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"example","version":"1.0.0"}}}'

Authentication

No authentication is required for the public API or MCP endpoints. They expose only information already published on prescientai.com and cannot access customer accounts, customer datasets, forecasts, or models. There are no API keys or OAuth scopes to request for this surface. See auth.md for the machine-readable policy.

Local sandbox

Use the live public sandbox at https://prescientai.com/sandbox/api/v1 to verify connectivity without credentials or customer data. For local development, run npm install and npm run dev, then use http://localhost:4321. The public API is read-only, so sandbox, local, and production calls cannot mutate customer data.

Errors and rate limits

REST errors use one JSON shape with code, message, hint, and documentation_url. MCP errors follow JSON-RPC 2.0 and tool-level failures set isError: true. Responses include standard rate-limit fields for agent planning.

{
  "error": {
    "code": "missing_query",
    "message": "The q query parameter is required.",
    "hint": "Try ?q=marketing+mix+modeling",
    "documentation_url": "https://prescientai.com/developers#errors"
  }
}

Versioning and support

Stable REST operations live under /api/v1. Breaking changes move to a new major path; additive response fields may be released within v1. Prescient announces deprecations in this portal and the OpenAPI document at least 180 days before removal, and affected responses carry Deprecation, Sunset, and successor-version Link headers during that window. Read the complete API versioning and deprecation policy. The active date-based build version is returned in X-API-Version. For questions, use the contact page.

Prescient AI Developer Portal