# Prescient AI MCP servers

Prescient AI publishes two public Model Context Protocol servers using Streamable HTTP. No authentication or credentials are required because these servers expose public product and documentation information only.

## Product MCP server

- Endpoint: https://prescientai.com/mcp
- Server card: https://prescientai.com/.well-known/mcp/server-card.json
- Purpose: Check integrations and pricing, build product-evaluation plans, return the human-reviewed demo workflow, and create hypothetical channel-budget scenarios.
- Tools: list_prescient_integrations, get_prescient_pricing, get_demo_booking_url, create_media_measurement_plan, and create_budget_scenario.

## Documentation MCP server

- Endpoint: https://prescientai.com/mcp/docs
- Server card: https://prescientai.com/.well-known/mcp/docs-server-card.json
- Purpose: Search and retrieve public Prescient AI documentation and educational content.
- Tools: search_prescient_docs and get_prescient_content.

## Protocol support

Both endpoints support initialize, ping, tools/list, tools/call, resources/list, and resources/read over JSON-RPC 2.0. They also support the 2026-07-28 server/discover negotiation method. Responses are public and read-only; hypothetical planning output must not be represented as customer-specific model results.

## Start here

Read the [developer portal](https://prescientai.com/developers), the [agent instructions](https://prescientai.com/agent-instructions.md), or the [OpenAPI specification](https://prescientai.com/openapi.json) for related integration surfaces.