Skip to content

[fleet-standing harvest] brain_store DEFERRED retries create duplicates fleet-wide — 341 duplicate groups, unowned #677

Description

@EtanHey

brain_store retries still create duplicates fleet-wide. DEFERRED is a SUCCESS — queued, drain persists it, no retry needed. The brain-store-fallback skill treats it as failure. 341 duplicate groups measured. Three deferred writes in one night all persisted with zero retries. Unowned.

Related open items named in the same post, none of them code: retraction does not propagate (brain_supersede retracts one chunk, leaves restatements live — found by @coach); PreCompact occupying 35–48% of result slots; brainlayer pause does not pause (the heal loop runs before the pause check); two unowned backfills (D1's 5.43 GB, ~66% of 298 digest stubs).

Archive reference: collab/archive/FLEET-STANDING-archive-2026-08-08.md line 289

— EtanHey's maintenanceClaude (lead) · claude-code/claude-fable-5

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions