# Agentic Card — AI agent instructions

Agentic Card is a searchable index of AI agents (A2A, MCP, OpenAPI, portable manifests) with trust signals. Use the HTTP API and static manifest below; do not scrape the HTML UI as the source of truth.

## Base URL

Use the deployment origin (e.g. `https://your-domain.com`). All paths below are relative to that origin.

## Discovery manifest

- `GET /.well-known/agent.json` — Agentic Card MCP-style manifest (capabilities, skill ids, discovery metadata).

## Public API (JSON)

- `GET /api/search?q={keyword}` — Full-text search; query `q` is required.
- `GET /api/agents` — Paginated listing; supports filters and `sort` (e.g. `trust`), `per_page`.
- `GET /api/agents/{slug}` — Single agent detail.
- `GET /api/agents/{slug}/trust` — Trust score and breakdown.
- `GET /api/agents/{slug}/health` — Health / uptime signals where available.
- `GET /api/stats` — Aggregate counts and summaries.
- `GET /api/categories` — Categories with counts.

## Human docs

- `/api` — API documentation in the app.
- `/about` — Project context.

## Safety

Listings are community-sourced. Verify endpoints, auth, and behavior before granting access in production; trust scores are signals, not guarantees.