Skip to content

This release adds 2 notable features for engineering teams evaluating rollout.

✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai ai-agents ai-memory anthropic artificial-intelligence claude
+12 more
claude-agent-sdk claude-agents claude-code-plugin claude-skills codex embeddings long-term-memory memory-engine openclaw openclaw-skills postgresql llm

Summary

AI summary

Updates Summary by CodeRabbit, Test plan, and auto-pull across a mixed release.

Changes in this release

Feature Low

Adds `DeeplakeApi.knownTablesOrNull()` returning cached table list or null for untrusted lookups.

Adds `DeeplakeApi.knownTablesOrNull()` returning cached table list or null for untrusted lookups.

Source: llm_adapter@2026-05-27

Confidence: high

Feature Low

Introduces optional `tableExists` predicate in `renderContextBlock` and `runPull` to skip SELECT for absent tables.

Introduces optional `tableExists` predicate in `renderContextBlock` and `runPull` to skip SELECT for absent tables.

Source: llm_adapter@2026-05-27

Confidence: high

Feature Low

Wires new existence‑check logic into session‑start hooks and auto‑pull utilities.

Wires new existence‑check logic into session‑start hooks and auto‑pull utilities.

Source: llm_adapter@2026-05-27

Confidence: high

Feature Low

Adds test coverage for table‑existence validation across multiple features.

Adds test coverage for table‑existence validation across multiple features.

Source: llm_adapter@2026-05-27

Confidence: high

Bugfix Medium

Prevents SELECT errors for non‑existent tables during session start.

Prevents SELECT errors for non‑existent tables during session start.

Source: llm_adapter@2026-05-27

Confidence: high

Full changelog

Summary

SessionStart reads hivemind_rules / hivemind_goals (context renderer) and skills (auto-pull) on every start, relying on catching relation does not exist. The client handled it, but the SELECT still failed server-side, so pg-deeplake logged a 42P01 per read on every fresh/empty workspace — the single largest cluster in the prod error digest (~1,060/24h, dominated by hivemind_goals/hivemind_rules/skills).

This gates those reads on a trusted table list so we never issue the SELECT for a table that isn't there.

  • DeeplakeApi.knownTablesOrNull() — returns the cached table list, or null when the lookup is untrusted (lets callers tell an empty workspace apart from a failed fetch).
  • renderContextBlock and runPull take an optional tableExists predicate; when it reports a table absent, the SELECT is skipped (section/rows empty).
  • Wired from the read entrypoints: hooks/session-start.ts (claude), hooks/hermes, hooks/cursor, commands/context.ts, and skillify/auto-pull.ts. (codex doesn't render the block; it only auto-pulls, covered centrally.)

Fallback preserved: the existing isMissingTableError catch stays. When the list is unavailable (predicate omitted), behavior is unchanged — a transient lookup blip never drops a read of a table that exists. The pre-check only fast-skips when we have a confident list showing the table is absent.

listTables() is cached and already warmed by the write paths each session, so this is typically zero extra HTTP calls.

Scope: kills the bare-name reads (hivemind_rules/hivemind_goals/skills). Schema-qualified stragglers (default.sessions, etc.) come from SDK/shell paths and are out of scope.

Test plan

  • [x] npm run typecheck clean
  • [x] New regressions: renderer + pull skip the SELECT (assert no query fired) when tableExists reports absent, and still SELECT when present
  • [x] Updated DeeplakeApi test mocks with knownTablesOrNull()
  • [x] Touched-module suites green; full suite green except 5 pre-existing failures unrelated to this change (cli-index / install-consent install-consent flow, standalone-embed-client defaults — none import these modules)

Summary by CodeRabbit

  • New Features

    • Added table existence checking to prevent errors when querying non-existent database tables during initial setup.
  • Bug Fixes

    • Improved handling of missing database tables in context rendering and data synchronization to gracefully skip unavailable resources instead of failing.
  • Tests

    • Added test coverage for table existence validation logic across multiple features.

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track Hivemind turns agent traces into skills and shares with your team

Get notified when new releases ship.

Sign up free

About Hivemind turns agent traces into skills and shares with your team

All releases →

Related context

Earlier breaking changes

  • v0.7.52 Removes `hivemind tasks` CLI and related code surfaces.
  • v0.7.51 Removes `hivemind tasks` CLI and related code surfaces.
  • v0.7.19 Module name skilify replaced with skillify; affects all imports
  • v0.7.19 CLI command skilify removed; renamed to skillify without deprecation alias
  • v0.7.18 CLI subcommand renamed from `skilify` to `skillify`; no deprecation alias.

Beta — feedback welcome: [email protected]