Is your feature request related to a problem?
All knowledge base endpoints currently use a single bucket, making it difficult to perform a clean data flush for NGOs, resulting in offboarding issues.
Describe the solution you'd like
- Organize data into per-project-ID folders within the same bucket (following the audio setup).
- Implement a migration script for existing documents and a legacy fallback option in the code (check top level, then project level).
- Conduct testing on staging first.
- Defer the bring-your-own-bucket/credentials aspect until TAP.
Priority
- P2, September in-person ops/infra sprint.
Original issue
Context
For knowledge base enpoints, everything currently lands in a single bucket. If an NGO uses the product for two years and then requests a full data flush, that's very hard to honour cleanly — an offboarding liability.
Fix
- Same bucket, per-project-ID folders (audio already does this; documents may not).
- Provide a migration script for existing documents and a legacy fallback in code (check top level, then project level).
- Test on staging first.
- The bring-your-own-bucket/credentials half of this is deferred until TAP.
Priority
- P2, September in-person ops/infra sprint.
Is your feature request related to a problem?
All knowledge base endpoints currently use a single bucket, making it difficult to perform a clean data flush for NGOs, resulting in offboarding issues.
Describe the solution you'd like
Priority
Original issue
Context
For knowledge base enpoints, everything currently lands in a single bucket. If an NGO uses the product for two years and then requests a full data flush, that's very hard to honour cleanly — an offboarding liability.
Fix
Priority