Planned chunk_set migration for paper-kb backend

  • Day: 2026-05-19
  • Time: 11:10 to 11:25
  • Project: Dev
  • Workspace: WP 2: Operational
  • Status: In Progress
  • Priority: HIGH
  • Assignee: Matías Nehuen Iglesias
  • Tags: Chunk_Set, Paper-Kb, Architecture, Migration, Contracts, Backend

Description

Session Goal

Analyze the boundary between KB, paper-kb, OpenAlex, Chroma, and abstract-scroller, with a focus on making chunk_set the canonical runtime substrate for paper-kb before any repository split or broader storage refactor.

Key Activities

  • Reviewed multiple diagnostic and planning notes about multi-repo architecture, repository boundaries, and contract ownership.
  • Compared the documented chunk-bus / chunk_set contract with the current runtime serving path in paper-kb.
  • Identified that paper-kb still depends on legacy storage adapters and Chroma-facing paths, which conflicts with the intended canonical artifact flow.
  • Framed chunk_set as the shared integration seam and migration bridge across repositories.
  • Outlined a staged migration strategy:
    • define chunk_set.v1 in KB,
    • add a ChunkSetStorageAdapter to paper-kb,
    • keep Chroma as a derived index rather than the primary substrate,
    • defer OpenAlex and abstract-scroller bridges until the paper-kb path is stable.
  • Requested additional repository inspection guidance and prioritized file checks for backend, pipeline, and test areas to continue the diagnostic work.

Achievements

  • Clarified the architectural mismatch between the intended contract-driven design and the current serving implementation.
  • Established a boundary-first refactor direction centered on canonical chunk_set artifacts.
  • Defined the likely ownership split: KB owns the contract, paper-kb consumes and validates it, and downstream systems remain secondary until the bridge is stable.
  • Produced a concrete migration intent that favors small, safe PRs over a big-bang rewrite.

Pending Tasks

  • Inspect the remaining paper-kb backend, pipeline, and test files to confirm the actual serving path and adapter requirements.
  • Formalize chunk_set.v1 and validate schema/contract expectations.
  • Implement or prototype ChunkSetStorageAdapter in paper-kb.
  • Verify how Chroma can be demoted to a derived index without breaking current behavior.
  • Define the exact first migration step and the minimal smoke tests needed to de-risk the transition.

Evidence

  • source_file=2026-05-19.sessions.jsonl, line_number=3, event_count=0, session_id=a8872494031d99f6babf11464dd85575ce0ac79e3939efbfcb5841410f29970d
  • event_ids: []