Context
Reviewed a Stremio setup guide from a previous session and identified shallow research: a handful of WebSearches, then straight to writing. The guide contained multiple incorrect claims that deeper research would have caught (Chromecast casting broken in v5, Android app removed cast button, sideloading unnecessary for Google TV). Created ecosystem-wide guidance to prevent this pattern.
What Was Done
New Guidance: deep-research.md
- 6-step research workflow: scope, deep dive, gotcha search, verify, completeness check, write
- Minimum source diversity requirement (3+ categories)
- Explicit gotcha hunting (search for problems/issues before recommending)
- Cross-referencing key claims across 2+ independent sources
- Version/platform disambiguation
- Completeness audit checklist
- Added as ESSENTIAL rule #13 (auto-loaded at SessionStart)
Stremio Guide Rewrite
Rewrote the guide following the new methodology. 20+ searches and fetches across official docs, GitHub bug tracker, community forums, configuration pages, and troubleshooting resources.
Critical corrections found through deep research:
- Stremio v5 desktop does NOT support Chromecast casting (dev confirmed not implemented)
- New Android app 2.0+ removed the Chromecast button
- Stremio for Android TV is on Google Play Store since 2022 (sideloading unnecessary)
- Flix Streams is freemium (free tier: only 2 providers)
- Web client still supports Chromecast via Chrome Cast Framework
- Expert consensus: free/no-debrid mode has documented real costs (ISP letters, buffering)
Repos Touched
- agentGuidance (191fade) – new guidance file + ESSENTIAL rule #13
- llm-tasks (cb5278a) – rewritten Stremio guide
Key Takeaway
The gotcha search step alone (searching for Stremio Chromecast problems) surfaced the v5 casting regression, which completely invalidated the original guides primary recommendation. This is exactly the kind of thing the new guidance prevents.