134 0907aaff67 fix(agent): make detail replacement results survive page and server restarts ...

详情替换原来把整批生图挂在浏览器保持的对话流上，切换标签页、断网或连接中断后
结果就再也回不来，前端还会继续播放按时间推算的 96% 假进度。

现在改为服务器端后台批次：提交时先落任务，再按参考图逐张生成并把每张结果写进
项目，页面刷新、切页或服务重启后都能按任务状态恢复，不再依赖前端连接。

- 新增 /api/agents/detail-replacement/batches（提交/查询、幂等去重、重试上限）
- 前端只在服务端确认批次后显示处理中，卡片显示排队/生成/完成/失败真实状态
- 去掉假进度动画；没有产品输入图时提交前直接拦截
- 服务重启后自动恢复未完成批次，长批次不再被通用 30 分钟清理误判
- 单张失败只落到该卡片并继续后续图片，已完成的图片不重复生成和扣费
- 详情替换默认模型由持续 404 的 gpt-image-2.5-flare 改回 gpt-image-2.5

2026-09-26 02:39:11 +08:00

..

box-cutout

feat: add specialist direct executor

2026-03-07 16:59:16 +08:00

context-memory

fix(ai): preserve relational image references

2026-09-05 18:46:25 +08:00

extend

fix: bind image extension to Alibaba provider key

2026-09-10 12:48:27 +08:00

layer-split

chore: govern layer split logging

2026-05-27 13:12:05 +08:00

providers

fix(video-canvas): resume provider jobs and recover editing and 3d state

2026-09-17 15:01:02 +08:00

resize

feat: upgrade agent executor and canvas sync

2026-03-07 15:58:03 +08:00

rmbg

feat(ai): consolidate providers around VIP98K and FAL

2026-08-08 16:52:14 +08:00

runtime

fix: honor vip98k image contracts

2026-09-10 16:19:49 +08:00

specialists

fix canvas cutout intent routing

2026-07-07 21:39:56 +08:00

tools

fix(video-canvas): resume provider jobs and recover editing and 3d state

2026-09-17 15:01:02 +08:00

atlascloud-models.ts

feat: add AtlasCloud video model support

2026-07-23 10:52:29 +08:00

atlascloud-request.ts

fix: match video canvas models to real capabilities and remove simulated processing

2026-09-16 15:06:05 +08:00

auto-agent-graph.ts

feat(ai): consolidate providers around VIP98K and FAL

2026-08-08 16:52:14 +08:00

bananaimage-models.ts

feat: add BananaImage image provider

2026-07-13 23:04:54 +08:00

bltcy-image-models.ts

feat: add BLTCY Flux image providers

2026-04-18 10:29:30 +08:00

bltcy-video-models.ts

feat: add bltcy unified video provider

2026-04-23 18:24:39 +08:00

canvas-agent-routing.ts

feat: pass canvas slot context to agent planner

2026-06-23 14:03:02 +08:00

classifier.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

conversation-action-resolver.ts

fix: improve agent image intent handling

2026-05-11 20:31:17 +08:00

conversation-memory-composer.ts

feat: strengthen agent runtime boundaries

2026-03-16 11:45:00 +08:00

conversation-state-protocol-events.ts

feat: unify agent workflow protocol

2026-03-13 16:16:09 +08:00

conversation-state-protocol.ts

fix: preserve image follow-up context

2026-03-16 15:50:01 +08:00

conversation-thread-state-builder.ts

Update conversation memory and canvas persistence

2026-03-13 12:31:13 +08:00

detail-agent-kimi-profile.ts

feat: refine canvas workflows and unify application UI

2026-07-10 12:57:01 +08:00

detail-agent-model-config.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

detail-agent-model-runtime.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

detail-replacement-image-model-config.ts

fix(agent): make detail replacement results survive page and server restarts

2026-09-26 02:39:11 +08:00

detail-replacement-inputs.ts

fix(agent): make detail replacement results survive page and server restarts

2026-09-26 02:39:11 +08:00

detail-replacement-vision-model-config.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

detail-replacement-vision-runtime.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

dynamic-provider.ts

feat(ai): add VIP98K provider foundation on 48 baseline

2026-08-08 16:52:14 +08:00

entitlements.ts

fix: support agent admin chat entitlements

2026-03-24 16:53:17 +08:00

fal-image-models.ts

merge: integrate canvas and UI refinements

2026-07-10 13:18:21 +08:00

grsai-gpt-image-sizes.ts

fix image providers and model deletion

2026-06-11 19:35:09 +08:00

image-edit-subagent.ts

feat: route canvas agent local edits

2026-06-24 00:17:03 +08:00

image-generation-subagent.ts

fix: improve agent image intent handling

2026-05-11 20:31:17 +08:00

image-model-pricing.ts

feat: add image size credit multipliers

2026-05-22 20:49:14 +08:00

image-size-constraints.ts

Normalize image output dimensions

2026-06-13 13:33:02 +08:00

image-workflow-memory.ts

Update conversation memory and canvas persistence

2026-03-13 12:31:13 +08:00

krea-models.ts

fix: refresh krea image provider integration

2026-05-30 12:53:18 +08:00

model-classification.ts

feat(ai): support Qwen Image 3.0

2026-08-26 16:10:47 +08:00

model-compat.ts

fix: stabilize canvas agent mode for kimi

2026-04-05 01:28:28 +08:00

model-frontend-availability.ts

fix: polish style pro model visibility and billing

2026-05-20 14:26:24 +08:00

model-router.ts

feat(ai): add VIP98K provider foundation on 48 baseline

2026-08-08 16:52:14 +08:00

model-runtime.ts

fix: keep ruxa image modelId from configuration without remap

2026-06-08 21:22:30 +08:00

model-visibility.ts

fix: restore production model availability helpers

2026-08-20 19:28:51 +08:00

models.mock.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

models.test.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

models.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

prompt-memory.ts

Update conversation memory and canvas persistence

2026-03-13 12:31:13 +08:00

prompt-optimizer.ts

feat(ai): consolidate providers around VIP98K and FAL

2026-08-08 16:52:14 +08:00

prompts.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

provider-defaults.ts

feat(ai): add VIP98K provider foundation on 48 baseline

2026-08-08 16:52:14 +08:00

providers.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00

qwen3-vl.ts

fix(canvas): decouple text detection from image provider

2026-08-25 14:03:51 +08:00

qwen25-vl.ts

fix(canvas): decouple text detection from image provider

2026-08-25 14:03:51 +08:00

reference-image-capability.ts

fix: stabilize ai wash model handling

2026-07-01 00:30:56 +08:00

replacement-layout-lock.ts

fix: preserve detail replacement product geometry

2026-09-09 19:50:03 +08:00

replacement-product-identity.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

runway-video-models.ts

fix: align runway video params with model capabilities

2026-03-21 01:15:24 +08:00

ruxa-image-models.ts

fix(ruxa): use configured task input mode for image models

2026-06-09 17:11:10 +08:00

skill-router.ts

feat: route canvas agent local edits

2026-06-24 00:17:03 +08:00

smart-analyzer.ts

feat(ai): consolidate providers around VIP98K and FAL

2026-08-08 16:52:14 +08:00

snapgen-image-models.ts

fix: infer SnapGen auto aspect ratio

2026-06-24 11:35:41 +08:00

tools-registry.ts

feat: route canvas agent local edits

2026-06-24 00:17:03 +08:00

video-model-config.ts

feat: show generation point estimates

2026-04-18 11:58:37 +08:00

video-tool-capability.ts

feat: enhance workflow controls and media settings

2026-04-20 14:39:41 +08:00

vip98k-models.ts

fix: match video canvas models to real capabilities and remove simulated processing

2026-09-16 15:06:05 +08:00

vip98k-reference-image.ts

fix: verify VIP98K sizes and recover replacement results

2026-09-11 14:52:38 +08:00

vip98k-reference-staging.ts

fix: verify VIP98K sizes and recover replacement results

2026-09-11 14:52:38 +08:00

vip98k-result-delivery.ts

fix: update vip98k image task integration

2026-09-07 00:22:45 +08:00

vip98k-runtime-config.ts

perf: cache vip98k reference staging

2026-09-07 20:50:20 +08:00

vip98k-runtime.ts

feat(ai): consolidate providers around VIP98K and FAL

2026-08-08 16:52:14 +08:00

vip98k-task-invocation.ts

fix: match video canvas models to real capabilities and remove simulated processing

2026-09-16 15:06:05 +08:00

vision-inline.ts

fix: stabilize detail replacement generation

2026-09-11 11:36:53 +08:00

visionary-models.ts

feat: add Visionary image provider

2026-07-14 14:12:27 +08:00

volcano-video-models.ts

Initial commit - clean repository

2025-12-17 20:15:18 +08:00