Recovery tier index
Open this after web browse returns typed unsupported content or NO_TEXT_BLOCKS.
Choose one route; do not silently escalate through several services.
| Need | Load | Boundary |
|---|---|---|
| Direct PDF or known paper identifier | Tier 0 | Download only an authorized public URL; extract locally. |
| DOI, arXiv, PMID, or free structured API | Tier 1 | Verify the current vendor API and access policy. |
| Ordinary static article/blog/documentation | Tier 1.5 | No JavaScript or login. |
| Tables, lists, metadata, known selectors | Tier 2 | Requests + BeautifulSoup; no browser interaction. |
| JS-rendered public page | Tier 3 | Verify installed API; use the documented Nature/Springer wait mode. |
| Authorized external page-to-markdown service | Tier 4 | Check current account, quota, price, and access policy first. |
| Discovery through an external search vendor | Tier 5 | Explicit selection only; not a built-in web engine. |
For forms, login, uploads, SPA interaction, or human verification use
agent-native-browser, not a tier or web browse.
Before executing a retained helper or trusting an asset, read its known limitations.