# LRRK Published Archive Agent service

This service exposes three bounded, read-only operations over published LRRK archive material.

- Agent Card: <https://agents.lrrk.ai/.well-known/agent-card.json>
- AI Catalog: <https://agents.lrrk.ai/.well-known/ai-catalog.json>
- MCP Server Card: <https://agents.lrrk.ai/mcp/server-card>
- Skills index: <https://agents.lrrk.ai/skills.md>
- This documentation: <https://agents.lrrk.ai/docs/agent-service.md>

## Unauthenticated discovery

The Agent Card, AI Catalog, MCP Server Card, skills index, this documentation, and health status are public discovery documents. The MCP Server Card describes only the existing authenticated Streamable HTTP endpoint; it is not an authorization grant. Public documents use `X-Content-Type-Options: nosniff` and `Referrer-Policy: no-referrer`; Markdown also uses a restrictive Content-Security-Policy.

## GitHub-backed OAuth authorization

Using the A2A interface at <https://agents.lrrk.ai/a2a> requires OAuth authorization. GitHub identity and local user consent support that authorization; the only requested service scope is `archive.read`. Discovery does not authorize archive access.

## Read-only operations

- `list_published_items` — **List published LRRK archive items**: List a bounded page of published archive catalog metadata.
- `get_published_item` — **Get one published LRRK archive item**: Retrieve one published Markdown item by its exact catalog ID.
- `search_published_catalog` — **Search the published LRRK archive catalog**: Search bounded published catalog metadata without reading item bodies.