Console v2 dev: new-user email verification is a dead end (Resend unconfigured), blocking domain-capture org auto-join
Degenerate estimators in tag-graph analytics: label-propagation hairball needs a lift-threshold sweep, and per-session HHI over rarest-tags is scattered by construction
Pinned/restored browser tabs poison session-level "gateway" and "revisit" analyses — use a [REDACTED] and annotate opener share
Share-ratio metrics: eps-padded denominators and mismatched normalizations both fabricate signal — gate support and normalize both sides from the same pool
Multiple inline SVGs on one page: duplicate clipPath ids silently clip unrelated charts (first id wins document-wide)
Lift-ranked co-occurrence "surprising pairs" boards saturate with ceiling artifacts — support floors alone don't fix it
sklearn KMeans cluster_centers_ is already dense — calling .toarray() raises AttributeError
systemd-run with Python: log file stays empty due to stdout buffering
Headless Playwright mermaid verification: pw-run scripts must live under $HOME + launch system Chrome
[REDACTED] authoritative fast-path disables PII/NER scrubbing — unsafe for repo-mined text
Claude Code Stop hook fires per completed turn; SIGKILL mid-generation fires NEITHER Stop nor SessionEnd — bind per-turn side effects to Stop, not SessionEnd
claude-opus-4-8 streams EMPTY thinking deltas via Claude Code — "missing thoughts" is the model, not your display pipeline
burst() entry nodes: selectDiverseSeeds' type-diversity bonus silently evicts higher-scoring seeds from the top-5
Chat turns processing queued (mid-turn) messages silently vanish from [REDACTED] — parallel code path missed by earlier persist fix
SOMNUS.md spec drift: absolute "cannot autonomously create" claim was only true of one subsystem
Boolean divergence flag manufactured false confidence when shadow-compare never armed
Cypher NOT ... STARTS WITH on a nullable property silently disables a guard clause (three-valued logic)
Neo4j migration runner: dry-run mode via paired read-only count cyphers + MERGE-collision data-loss detection
Latency regression misattributed to LLM retry-broadening; real cause was connection-pool contention