Expose
MCP server endpoints
Make your organizational context available to any AI agent. Expose creates MCP endpoints that work with Cursor, Claude, and any MCP-compatible client.
MCP Native
Full Model Context Protocol support. Drop-in integration with Claude, Cursor, and MCP-compatible agents.
Context Windows
Dynamically size context for the target model. Optimize for GPT-4, Claude, Gemini, or custom LLMs.
Tool Definitions
Expose your context as tools. Agents can search, lookup, and act on your organizational knowledge.
Rate Limiting
Per-agent quotas and rate limits. Control costs and prevent abuse.
Context Caching
Intelligent caching reduces latency and API costs for repeated queries.
Observability
Full visibility into what context each agent receives. Debug and audit agent behavior.
// Note: MCP endpoints are configured via the Metalogue dashboard
// or the /v1/expose API. Connect from your AI tool:
// In Cursor settings or Claude config:
// mcp://mcp.metalogue.xyz/endpoints/your-org
// Or query programmatically:
import { MetalogueClient } from '@metalogue/sdk';
const client = new MetalogueClient({
apiKey: process.env.METALOGUE_API_KEY
});
// The same query API works for both direct calls
// and MCP-exposed context
const response = await client.query(
'How do we handle auth in the mobile app?',
{ limit: 5, sources: ['notion', 'github'] }
);