chore: publish new package versions

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@electric-ax/agents@0.6.3

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
  • @electric-ax/agents-runtime@0.6.3

@electric-ax/agents-runtime@0.6.3

Patch Changes

• 886c423: Context compaction for the agents runtime. Modelled on OpenAI Codex's
  summarization but adapted to the event-sourced timeline (a context_inserted
  checkpoint placed at a stored watermark, so reconstruction folds older messages
  into a summary):

  • A context-window usage gauge (cache-inclusive context_input_tokens +
    context_window persisted per step) and <token_budget> notices injected at
    25 / 50 / 75% usage.
  • Oversized tool-output truncation, and a synchronous mid-turn compaction floor
    at the 90% hard ceiling (runs before every model step via the adapter's
    transformContext hook).
  • Non-blocking background (turn-end) compaction that starts at 85%: a detached
    summarize whose checkpoint is applied at the next turn's start, or immediately
    if it finishes while idle. Each generation uses a watermark-unique checkpoint
    id so a new run can't supersede a prior completed one. Summarize calls are
    bounded by a hard timeout.
  • UI: a "Compacting…" indicator (blocking vs. background) and a collapsible
    "Context compacted" entry in the conversation timeline.

  Thresholds are env-tunable (ELECTRIC_AGENTS_COMPACT_CEILING,
  ELECTRIC_AGENTS_COMPACT_BG_CEILING).

• 7139c53: Context-usage indicator: a circular ring gauge plus a hover breakdown popover
  showing how the prompt decomposes across system prompt, tools, messages, and
  free space (à la Claude Code's /context). The runtime now persists an
  approximate context_breakdown of the stable request parts (system + tools) on
  each step alongside the cache-inclusive total, and exports new
  computeContextBreakdown / parseContextBreakdown helpers from the client
  entry; the UI derives the "messages" bucket as the real remainder so the
  segments always sum to the gauge.

@electric-ax/agents-server@0.6.3

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
• Updated dependencies [4f604bb]
  • @electric-ax/agents-runtime@0.6.3
  • @electric-sql/client@1.5.22

electric-ax@0.2.23

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
• Updated dependencies [4f604bb]
  • @electric-ax/agents-runtime@0.6.3
  • @electric-sql/client@1.5.22
  • @electric-ax/agents@0.6.3

@electric-sql/experimental@6.0.22

Patch Changes

• Updated dependencies [4f604bb]
  • @electric-sql/client@1.5.22

@electric-sql/react@1.0.51

Patch Changes

• Updated dependencies [4f604bb]
  • @electric-sql/client@1.5.22

@electric-sql/client@1.5.22

Patch Changes

• 4f604bb: Fix ShapeStream wake reconnects in runtimes that do not preserve AbortSignal.reason, such as some React Native fetch/AbortController implementations.

@electric-sql/y-electric@0.1.48

Patch Changes

• Updated dependencies [4f604bb]
  • @electric-sql/client@1.5.22

@electric-ax/example-agents-chat-starter@0.1.14

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
  • @electric-ax/agents-runtime@0.6.3

@electric-ax/example-agents-walkthrough@0.1.11

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
  • @electric-ax/agents-runtime@0.6.3

@electric-ax/example-deep-survey@0.1.30

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
  • @electric-ax/agents-runtime@0.6.3

expo-db-electric-starter@1.0.23

Patch Changes

• Updated dependencies [4f604bb]
  • @electric-sql/client@1.5.22

@electric-ax/agents-mobile@0.6.3

Patch Changes

• Updated dependencies [886c423]
• Updated dependencies [7139c53]
  • @electric-ax/agents-runtime@0.6.3
  • @electric-ax/agents-server-ui@0.6.3

@electric-ax/agents-server-ui@0.6.3

Patch Changes

• 886c423: Context compaction for the agents runtime. Modelled on OpenAI Codex's
  summarization but adapted to the event-sourced timeline (a context_inserted
  checkpoint placed at a stored watermark, so reconstruction folds older messages
  into a summary):

  • A context-window usage gauge (cache-inclusive context_input_tokens +
    context_window persisted per step) and <token_budget> notices injected at
    25 / 50 / 75% usage.
  • Oversized tool-output truncation, and a synchronous mid-turn compaction floor
    at the 90% hard ceiling (runs before every model step via the adapter's
    transformContext hook).
  • Non-blocking background (turn-end) compaction that starts at 85%: a detached
    summarize whose checkpoint is applied at the next turn's start, or immediately
    if it finishes while idle. Each generation uses a watermark-unique checkpoint
    id so a new run can't supersede a prior completed one. Summarize calls are
    bounded by a hard timeout.
  • UI: a "Compacting…" indicator (blocking vs. background) and a collapsible
    "Context compacted" entry in the conversation timeline.

  Thresholds are env-tunable (ELECTRIC_AGENTS_COMPACT_CEILING,
  ELECTRIC_AGENTS_COMPACT_BG_CEILING).

• 7139c53: Context-usage indicator: a circular ring gauge plus a hover breakdown popover
  showing how the prompt decomposes across system prompt, tools, messages, and
  free space (à la Claude Code's /context). The runtime now persists an
  approximate context_breakdown of the stable request parts (system + tools) on
  each step alongside the cache-inclusive total, and exports new
  computeContextBreakdown / parseContextBreakdown helpers from the client
  entry; the UI derives the "messages" bucket as the real remainder so the
  segments always sum to the gauge.

• Updated dependencies [886c423]

• Updated dependencies [7139c53]

  • @electric-ax/agents-runtime@0.6.3

@core/sync-service@1.7.4

Patch Changes

• e6c3767: Emit shape_snapshot.execute_for_shape and shape_snapshot.query_fn spans for move_in_query operations. Previously these spans were silently dropped because the spawned move-in task had no parent span context, and with_child_span skips emission when no parent exists. A shape_snapshot.move_in_task root span is now opened inside the task (mirroring shape_snapshot.create_snapshot_task for initial snapshots) so the existing child spans are emitted with shape.query_reason="move_in_query".
• 44eafb6: Fix subquery shapes diverging from Postgres after a server restart. Shapes
  whose where clause contains a subquery (e.g. id IN (SELECT ... WHERE active))
  could return stale results or a 409 must-refetch once the stack restarted and
  restored from disk. Two causes are addressed: the dependency materializer now
  replays the full persisted history (snapshot + main log) on startup instead of
  just the first chunk, and dependent (outer) subquery consumers are eagerly
  restarted during restore so their materializer subscription is re-established
  before live events flow.
• 121dbe6: Reduce the memory footprint of the subquery filter index. The per-value ETS
  rows that back subquery routing and exact membership repeated several boxed
  terms in their keys — the shape handle, make_ref/0 condition references, the
  canonical ["$sublink", "<dep>"] subquery ref, and the {condition_id, field}
  node id. Each of those is now a compact integer id, leaving only the actual
  typed value boxed. This cuts the cost per seeded subquery membership value by
  ~51% (e.g. ~568 MiB → ~275 MiB for 1,000,000 seeded values) with no change to
  routing behaviour.
• cf476c1: Reset replication metadata caches when shapes are purged on a new or temporary
  replication slot. Previously, when Electric purged all shapes after the
  replication slot was (re)created — e.g. a temporary slot recreated on restart
  with CLEANUP_REPLICATION_SLOTS_ON_SHUTDOWN=true — the persisted relation
  tracking (tracked_relations) and the inspector cache (ets_inspector_state)
  survived and were read back on startup, even though WAL continuity had been
  lost. These caches are now reset as part of the same purge, so no stale
  pre-restart metadata is reused.

[netlify]

✅ Deploy Preview for electric-next ready!

Name	Link
🔨 Latest commit	5be4d94
🔍 Latest deploy log	https://app.netlify.com/projects/electric-next/deploys/6a3d51858e3e730008180cda
😎 Deploy Preview	https://deploy-preview-4645--electric-next.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.