总工
a175fe57dd
HEL-873 r2: merge PR #22 (visual investigation, c3039343)
2026-10-26 23:55:13 +08:00
施工员 and multica-agent
c3039343ca
HEL-883: audit deployed visuals and mount canvas lightbox regressions
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-26 23:04:14 +08:00
总工
2ad5d33c72
HEL-873: merge PR #21 (library delete reasons, 5163230)
2026-10-26 21:17:19 +08:00
总工 and multica-agent
e81f191e64
HEL-873: merge PR #20 (reference preview, e08b32c)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-26 21:17:15 +08:00
施工员2号 and multica-agent
5163230433
fix(HEL-872): library delete reports real kept reasons, not a merged vague one
...
List only excludes project-data refs, but delete also kept canvas-card
files with the same 'referenced' reason the UI merged into 被项目使用或
无法删除 — every fresh playground output (materialized as a canvas card)
looked undeletable from 素材库 with a wrong explanation.
Backend: split the kept reason into referenced (project data) vs
on_canvas (live canvas card); protections unchanged.
Frontend: per-reason kept summary with counts and an actionable hint,
error toast names failure and stays retryable, readable API error text.
E2E: isolated real-page regression (cancel/single/batch/retry) +
backend tests for the on_canvas reason.
Co-authored-by: multica-agent <github@multica.ai >
2026-10-26 20:51:03 +08:00
施工员 and multica-agent
e08b32c064
HEL-871: preview existing reference media from composer slots
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-26 20:32:27 +08:00
leefer
344fede198
Merge pull request 'HEL-850: 仅对翻译和润色关闭模型思考 (reasoning_effort none)' ( #19 ) from agent/2/9c4d105d16eb into leefer/hel-731-studio-project
2026-10-25 22:04:29 +08:00
leefer
a8efdd3a66
HEL-853: fix Chinese render approval and keep refusal priority (PR #18 , reviewed head 49dc71b7)
2026-10-25 21:36:25 +08:00
leefer
796dff27ad
HEL-849: unify agent composer select interactions (PR #17 , reviewed head e5356bd)
2026-10-25 21:10:26 +08:00
施工员2号 and multica-agent
4cda754eab
feat(HEL-850): disable thinking for translate/polish via reasoning_effort none
...
LM Studio's OpenAI-compat endpoint silently ignores chat_template_kwargs
(probed live); reasoning_effort "none" verifiably removes reasoning tokens
on both streaming and blocking paths. Pass it through the existing
LLMClient extra_body channel for translate/polish first and correction
requests only; all other callers keep their default policy.
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 20:47:12 +08:00
leefer and multica-agent
49dc71b727
HEL-853: recognize Chinese render approvals and prioritize refusals
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 20:43:42 +08:00
leefer and multica-agent
e5356bdc3c
fix(HEL-849): unify agent composer select interactions
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 20:26:23 +08:00
leefer
410eadea15
HEL-845: 恢复素材库日间选中按钮可读性 (PR #16 , reviewed head bc2a70a)
2026-10-25 02:27:11 +08:00
leefer and multica-agent
bc2a70ac5f
test(HEL-845): wait for settled before and after colors
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 02:13:53 +08:00
leefer and multica-agent
c1d4feb4a8
fix(HEL-845): restore readable library day selection tokens
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 02:12:57 +08:00
leefer
0ffa8d00bb
HEL-842+844: restore approved controls and translate weighted operands (PR #15 , reviewed head 13ff441)
2026-10-25 01:50:52 +08:00
leefer and multica-agent
13ff441378
HEL-844: translate weighted operands with verified syntax identities
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 01:16:06 +08:00
leefer and multica-agent
6e7760e304
HEL-842: report complete comparison manifest count
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 00:36:58 +08:00
leefer and multica-agent
29a3c29fa3
HEL-842: reload offline prototype when capturing model states
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 00:36:43 +08:00
leefer and multica-agent
27613336ff
HEL-842: restore approved library and model connection controls
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-25 00:35:31 +08:00
leefer
7fbcfad29e
Merge pull request 'HEL-840: 落地第二版工作室、素材库、模型分工与提示词工具' ( #14 ) from agent/hel-840-v2 into leefer/hel-731-studio-project
2026-10-25 00:16:12 +08:00
leefer and multica-agent
7ddd1d32d0
fix: complete approved layout and prompt cancellation safeguards
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:42:46 +08:00
leefer and multica-agent
4f4a75b3da
HEL-840: add isolated end-to-end evidence and reproducible review kit
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer and multica-agent
5bb859f809
HEL-840: unify connection details and prompt model assignments
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer and multica-agent
914fde7655
HEL-840: prioritize projects in studio and share header styling
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer and multica-agent
14a03c137d
HEL-840: implement approved three-column media library
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer and multica-agent
3bd80905ac
HEL-840: share preview and protected replacement across prompt fields
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer and multica-agent
cebe4b9326
HEL-840: add cancellable prompt transforms and independent model roles
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 23:10:07 +08:00
leefer
415fb78236
Merge pull request 'HEL-836: 按最初定版还原创作台并修复双环圆点' ( #13 ) from fix/hel-836-original-desk-orbits into leefer/hel-731-studio-project
2026-10-24 21:33:06 +08:00
leefer and multica-agent
15db5eb4ae
HEL-836 restore original creative desk and orbit animation
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-24 21:05:08 +08:00
leefer
a45b6c94f9
HEL-733/813/818 既定动效与已批准四主题画布 ( #12 )
...
Merge PR #12 : HEL-733/806/809/810/813/818 final wrap-up (reviewed head 37bd8d85 , HEL-819 independent review passed)
2026-10-24 00:55:36 +08:00
leefer and multica-agent
37bd8d85b8
test: map canvas viewport and minimap state in comparisons
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 18:54:50 +08:00
leefer and multica-agent
32d0297564
fix: resolve canvas viewfinder color in its scene
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 18:49:45 +08:00
leefer and multica-agent
f300927546
feat: apply approved four-theme canvas design (HEL-818)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 18:46:06 +08:00
leefer and multica-agent
fe4759b788
feat: implement approved restrained motion and state audit (HEL-813)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 18:00:46 +08:00
leefer and multica-agent
e2f139806f
fix(i18n): translate Build Scene suggestion labels (HEL-810)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 15:23:58 +08:00
leefer and multica-agent
b6ded3942b
fix(workflows): derive honest media metadata for single imports (HEL-809)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 15:08:19 +08:00
leefer and multica-agent
cb410fbdea
test: package HEL-806 evidence and verify genuine empty state
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 13:36:32 +08:00
leefer and multica-agent
6d6552e011
fix: complete HEL-806 audit evidence and shared visual corrections
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 13:28:29 +08:00
leefer and multica-agent
1f6c2d2a6b
HEL-733: align approved motion and audit remaining sample gaps
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-23 12:43:05 +08:00
leefer
2952af42b6
Merge pull request 'HEL-732: 智能体成果、项目草稿与样本视觉修复' ( #11 ) from feat/hel-732-agent-results into leefer/hel-731-studio-project
...
HEL-804: merge reviewed B4 agent results (HEL-732/785/786/788/791/792 chain; user approved 6 sample deviations on HEL-801)
2026-10-23 11:50:01 +08:00
leefer and multica-agent
ea3a1c6f06
fix(agent): retain daylight input focus border from sample
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 20:22:02 +08:00
leefer and multica-agent
e2b69aff16
fix(agent): preserve home title and media label visibility
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 20:18:37 +08:00
leefer and multica-agent
cb35d9819a
fix(agent): align header, tool cards and real message presentation
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 20:10:45 +08:00
leefer and multica-agent
afbf50447a
fix(agent): match studio controls, thumbnail ratio and sample digits
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 18:53:44 +08:00
leefer and multica-agent
ab48895b99
fix: regenerate SSE contract vocabulary (HEL-785)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 17:44:09 +08:00
leefer and multica-agent
dca216e2c7
test: align agent comparison content with real task events (HEL-732)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 16:09:33 +08:00
leefer and multica-agent
37cb6ec209
fix: keep agent task input reachable at small viewports (HEL-732)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 16:03:08 +08:00
leefer and multica-agent
31e5906462
feat: persist agent deliverables and explicit project draft handoff (HEL-732)
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 12:55:57 +08:00
leefer and multica-agent
8d7e58645c
fix: align B3 small button radii and primary disabled states
...
Co-authored-by: multica-agent <github@multica.ai >
2026-10-22 11:00:53 +08:00