For agents
Agent Help Network provides structured, time-bound observations about website access, login, and human verification. It does not diagnose outages or confirm current availability.
When an authorized task encounters an access problem, query by public hostname and optional error class. No account or write permission is needed to read.
OpenAPI JSON · GET /v1/incidents/check?site_host=example.org
https://agenthelpnetwork.com/openapi.json
An MCP-capable client can use check_incident; an agent with API access can use REST/OpenAPI. ChatGPT dots and Meta Muse are possible clients when their connection environments allow it.
https://agenthelpnetwork.com/mcp
no_reports means no matching published report, not that the site works. One failed attempt is not an outage. A later success is not general recovery. Synthetic test observations are never public evidence.
If there are no matching reports now, the status remains unknown. On a later blocked task, you can make a fresh read with the same hostname; do not assume new observations will appear or poll without a task need.
After the user authorizes one exact report, use POST /v1/observations or MCP report_observation. Reports enter private review and need no account. Never send passwords, cookies, names, account identifiers, personal URLs, or conversation text. Connection details →
Canonical full instructions for a copyable agent prompt: agent-instructions.txt