Skip to content

TODO

This page tracks near-term product and platform work that is not fully implemented yet.

Plan-Linked Gaps (Dummy/Stub Implementations)

  • [ ] Replace mock memory feed backend with real storage-backed pagination.
  • Plan: docs/plans/memory-feed.md
  • Current placeholder: main/server/api/memories/feed/app.py builds data via _build_mock_memories() with picsum URLs.

  • Target: query persisted user memories, preserve existing cursor/limit contract.

  • [ ] Replace mock conversation feed backend with real conversation summaries.

  • Plan: docs/plans/conversation-feed.md
  • Current placeholder: main/server/api/conversations/feed/app.py builds data via _build_mock_conversations().

  • Target: source conversations from persistent conversation history/state store while preserving pagination semantics.

  • [ ] Implement real memory search backend.

  • Plan: docs/plans/memory-search.md
  • Current placeholder: main/server/api/memories/search/app.py always raises RuntimeError("Search is not set up").

  • Target: support contract filters and return ranked results from actual indexed memory data.

  • [ ] Replace deterministic placeholder embeddings in ingest pipeline with model-based embeddings.

  • Plan: docs/plans/memory-upload-flow.md
  • Current placeholder: main/server/memories/pipeline/ingest/app.py _generateEmbeddingVector() uses deterministic hash-based vectors.
  • Target: integrate production embedding provider and persist true semantic vectors.

Product TODOs

  • [ ] Add support for text-based memories for non-real-world memory types (projects, ideas, notes, and related knowledge work).

  • Scope: memory create/read/search contracts plus UI flows.

  • [ ] Create a model/service to score whether a photo is "instagram worthy" and set highlight=true when appropriate.

  • Scope: scoring pipeline + persistence + feed UI behavior to render highlights as large tiles.