GenerAI / worldmonitor /public /agents.md
amogaddy's picture
Integra World Monitor (AGPL-3.0, self-hosted) nello Space: pagina, menu, e arricchimento notizie per la AI (part 3)
bd5a6ce verified
|
Raw
History Blame Contribute Delete
5.4 kB

World Monitor — Agent Guide

How AI agents should work with worldmonitor.app: machine surfaces, authentication, crawl policy, rate limits, and discovery endpoints. Prefer the structured surfaces below over scraping the HTML dashboard — the dashboard is a WebGL SPA and yields nothing useful to a text parser.

World Monitor is a real-time global intelligence dashboard: 500+ news feeds, 56 map layer types, country risk/resilience scores, AI briefs, forecasts, and market/supply-chain correlation, served as machine-readable JSON with documented methodology and provenance.

Machine surfaces (use these)

Authentication

  • Anonymous works for discovery endpoints, tools/list, and public data (world brief, product catalog, story pages).
  • API key: header X-WorldMonitor-Key: wm_<40-hex> for REST and MCP data calls — issue one at https://worldmonitor.app/pro. Full agent walkthrough: https://worldmonitor.app/auth.md
  • OAuth2 for MCP (scope=mcp), with dynamic client registration at /oauth/register. Details in auth.md.

Crawl & content-usage policy

  • robots.txt (https://www.worldmonitor.app/robots.txt): AI search/assistant agents (GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, Claude-SearchBot, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, DuckAssistBot, MistralAI-User) are explicitly allowed; bulk training-only scrapers (CCBot, Bytespider, anthropic-ai) are disallowed. /api/ is off-limits to crawlers except the allowlisted story/OG/llms.txt/product-catalog routes.
  • Content-Signal: ai-train=no, search=yes, ai-input=yes — declared as a robots.txt group directive and as an origin-wide HTTP response header. Search indexing and assistant grounding/citation are welcome; bulk model training is opted out.
  • User-Agent: always send a descriptive User-Agent (e.g. mytool/1.0 (+https://yoursite.example)). Default HTTP-library UAs (curl/*, python-requests/*, empty strings) may get a 403 from the edge firewall — a 403 does NOT mean the endpoint is missing; retry with a real UA.

Rate limits & plans

Support & escalation