# RealityPing

> Physical-world execution layer for AI agents. An agent orders a verified
> real-world observation; a human worker executes a structured mission;
> evidence is validated; a machine-readable result is returned over REST, MCP
> and A2A. Core loop: Quote -> Mission -> Result.

## For AI agents & developers (the machine surface — use this, not the landing page)
- Full agent guide (concept, MCP, autonomous onboarding): https://realityping.com/agents
- MCP server (Streamable HTTP): https://api.realityping.com/api/v1/mcp
- OAuth 2.1 discovery: https://api.realityping.com/.well-known/oauth-protected-resource/api/v1/mcp
- Dynamic Client Registration: POST https://api.realityping.com/api/v1/oauth/register
- Developer portal (API keys, sandbox playground): https://app.realityping.com
- API health: https://api.realityping.com/health
- Tools: reality_get_quote, reality_execute_mission, reality_get_mission, reality_get_result
- Auth: API key (Authorization: Bearer rly_...) or OAuth client_credentials.
  Autonomous agents need no human consent screen.

## Autonomous onboarding (an agent creates its own account, no human UI)
1. POST /api/v1/console/sign-in {"email": "..."}                -> session_token
2. POST /api/v1/console/oauth-clients (Bearer session_token)    -> client_id + client_secret
3. POST /api/v1/oauth/token (Basic client_id:secret,
   grant_type=client_credentials, resource=.../api/v1/mcp)      -> access_token (rly_oat_)
4. POST /api/v1/mcp (Bearer access_token)                       -> call tools
Sandbox: use API key rly_test_sandbox and place alias REALITY_SANDBOX_PARIS_STORE.

## For humans (the website)
- realityping.com recruits community members ("workers") who complete simple
  real-world missions for pay. The worker mobile app is in publication.