# YorpHives

> YorpHives is an open capability network for people and AI agents. Discover useful capabilities, inspect their contracts, invoke them, and compose them into workflows.

## Canonical discovery
- Capability catalog: https://yorphives-2-preview.pages.dev/api/v1/capabilities
- Machine-readable manifest: https://yorphives-2-preview.pages.dev/.well-known/yorphives-capabilities.json
- OpenAPI 3.1: https://yorphives-2-preview.pages.dev/openapi.json
- MCP Streamable HTTP: https://yorphives-2-preview.pages.dev/api/mcp

## Execution
POST https://yorphives-2-preview.pages.dev/api/v1/execute
Content-Type: application/json

{
  "capability": "text.analyze",
  "version": "1.0.0",
  "input": { "text": "Text to analyze" }
}

## Current public capability
- text.analyze@1.0.0 — deterministic text metrics and frequent terms.
- Pricing: free.
- Authentication: none.
- Input: plain text, maximum 20,000 characters.
- Retention: none; execution is in-memory.
- External side effects: none.

## MCP
- Endpoint: https://yorphives-2-preview.pages.dev/api/mcp
- Transport: Streamable HTTP
- Tool: text_analyze
- Discover tools with tools/list before invocation.

## Trust and safety
- Read the catalog before executing a capability.
- Do not send secrets or sensitive personal data to public capabilities.
- Pricing, authorization, limits, and data handling are declared in capability metadata.
- Errors use structured machine-readable codes and request IDs.

## Status
Early foundation. Capabilities marked available are implemented in the current deployment; roadmap items are not executable until explicitly published in the catalog.