f70c0214db
Align Assistant tool context and progress visibility
2026-06-06 04:02:07 +01:00
409cc9b836
Align Assistant prompt history recall
2026-06-06 03:48:47 +01:00
a7854bf341
Bound Assistant stream chunk stalls
2026-06-06 03:41:05 +01:00
ce4dc88ab7
Show Assistant model reasoning activity
2026-06-06 03:34:08 +01:00
729878b993
Show Assistant live wait duration
2026-06-06 03:28:19 +01:00
ab99112ad2
Show Assistant tool starts while streaming
2026-06-06 03:12:25 +01:00
3c3ca42c29
Align Assistant model selection with OpenCode
2026-06-06 02:55:59 +01:00
5398eeb795
Keep Assistant status on active tool
2026-06-06 02:27:46 +01:00
c858985c54
Add live Assistant active-turn status
2026-06-06 02:20:37 +01:00
8c7b14826d
Demote idle Assistant provider verification
2026-06-06 02:10:03 +01:00
f7cae1414e
Suppress redundant Assistant model override
2026-06-06 02:03:33 +01:00
30f22ee761
Normalize Assistant model route labels
2026-06-06 01:59:03 +01:00
7c583c0547
Answer Assistant inventory counts locally
2026-06-06 01:47:05 +01:00
16edfb44d3
Suppress Assistant tool prelude leaks
2026-06-06 01:20:37 +01:00
1f4477c27c
Show live Assistant tool progress
2026-06-06 01:12:01 +01:00
26d8baa6d8
Make Assistant progress live
2026-06-06 00:43:40 +01:00
e1a9b69862
Bound provider stream startup waits
2026-06-06 00:34:45 +01:00
26bdfe27df
Add same-model OpenRouter chat fallback
2026-06-06 00:26:02 +01:00
73b0635c21
Improve Assistant inventory and activity flow
2026-06-06 00:17:44 +01:00
9cdee482d5
Improve Assistant in-flight progress
2026-06-05 23:06:24 +01:00
2ab0a60e99
Speed up Assistant stream startup
2026-06-05 22:54:44 +01:00
0d2f2a3cc8
Anchor Assistant stream and OpenRouter default
2026-06-05 22:02:47 +01:00
5ef50f712d
Expose Assistant fallback model route
2026-06-05 21:29:41 +01:00
17c4c1c172
Scope Assistant fast inventory turns
2026-06-05 21:17:16 +01:00
206161b8ea
Route Assistant sends around failed providers
2026-06-05 20:48:54 +01:00
5e4b720d5f
Suppress duplicate Assistant thinking status
2026-06-05 20:40:59 +01:00
3d8a926acc
Keep Assistant tool-call leaks out of chat output
2026-06-05 20:29:24 +01:00
7067a51680
Make Assistant provider fallback resilient
2026-06-05 20:15:55 +01:00
9c06cda3ea
Bound Assistant stream provider startup
2026-06-05 18:14:14 +01:00
3b64c753c8
Show Assistant running tools inline
2026-06-05 18:04:35 +01:00
ff5eb8ac84
Add Assistant fallback route recovery
2026-06-05 17:53:04 +01:00
5462cefa19
Show Assistant response model routes
2026-06-05 17:45:25 +01:00
a54d643b7a
Make Assistant route recovery retry failed turns
2026-06-05 17:32:15 +01:00
7a05826ecd
Add direct Assistant provider-route recovery
2026-06-05 17:23:49 +01:00
67f896d93a
Improve Assistant failed-turn recovery and tool output
2026-06-05 17:11:43 +01:00
5ea51b4b25
Harden Assistant stream output and OpenRouter token budget
2026-06-05 16:30:07 +01:00
1f1ab45b32
Allow Assistant sends during provider warnings
2026-06-05 15:51:26 +01:00
064cd4bc80
Hide Assistant reasoning from message history
2026-06-05 15:43:06 +01:00
75706f8a2c
Harden Assistant output hygiene
2026-06-05 15:15:56 +01:00
bbef81274c
Show recent Assistant sessions in empty chat
2026-06-05 15:05:40 +01:00
5559075864
Hydrate Assistant loaded session tool events
2026-06-05 14:57:24 +01:00
e977db434f
Preserve Assistant retry request context
2026-06-05 14:50:53 +01:00
58769c3be9
Add Assistant prompt history recall
2026-06-05 14:41:41 +01:00
45f8757812
Add Assistant queued follow-up dock
2026-06-05 14:33:21 +01:00
af103d8b09
Queue Assistant follow-up sends
2026-06-05 14:23:24 +01:00
05d00a65a6
Hide raw Assistant tool-call output
2026-06-05 14:05:38 +01:00
df203b0ce6
Let Assistant send during provider checks
2026-06-05 13:38:00 +01:00
2b8ce06c03
metrics: drain freelist proportionally so metrics.db stops bloating
...
runRetention reclaimed only PRAGMA incremental_vacuum(5000) (~20MB) and
only when that hourly cycle deleted rows. On instances where an hourly
retention pass frees more than 5000 pages, the freelist grows net-positive
every cycle and the sqlite file bloats unboundedly (5GB+ of free pages over
~60MB of live data) even though row retention works. This is the documented
'50+ resources -> 5GB+' symptom the 5000-page batch was meant to fix.
Reclaim every cycle (so a pre-existing backlog drains even in a no-delete
hour) and size each reclaim to the current freelist, capped at 50000 pages
(~200MB) so a large backlog drains over several cycles without holding the
write lock for minutes. Skip the checkpoint when the freelist is empty so
steady-state WAL cadence is unchanged.
Surfaced by the demo server (pulse-relay): metrics.db had grown to 5.9GB
(5GB freelist), filling its 25GB disk and breaking the nightly backup.
Adds TestStoreRetentionReclaimsFreePages to the perf-and-scalability proof
set and documents the reclaim invariant in the subsystem contract.
2026-06-05 13:29:48 +01:00
2199423821
Gate Assistant sends on provider readiness
2026-06-05 13:28:34 +01:00
890b1d35ec
Clarify Assistant interruption state
2026-06-05 13:16:01 +01:00
rcourtman