Commit Graph

15 Commits

Author SHA1 Message Date
ywkj 6bc4e1d3b4 feat: image-only pipeline with LLM post-filter for category accuracy
register-skill-release / register (push) Successful in 18s Details
- Drop video-understanding flow (detect-video, video-analyzer.ts) — image
  search is the only path now since text/video keywords return broad results.
- Add container-aware frame selection: detect rack/holder products, restrict
  ranking to the earliest 40% of frames so empty/unboxing shots win over
  loaded ones (image search was matching shoes-on-rack instead of the rack).
- Switch container check from generateObject (silently fails on this model)
  to generateText with a YES/NO answer.
- Add post-filter step: send the snapshot + each result's pic_url to the
  vision model in batches, drop results whose category doesn't match the
  detected product description. Cuts 50 raw hits to ~10 same-type matches.
- When post-filter succeeds, sort by sales directly instead of running the
  keyword-intersection rerank, which was overriding good filtered results
  with broad keyword fallbacks.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-26 15:01:42 +08:00
ywkj e9e1f01728 docs: remove frame-extraction workflow from SKILL.md, keep video-direct approach only
register-skill-release / register (push) Successful in 20s Details
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-25 16:53:50 +08:00
ywkj db4735e54e feat: add detect-video command using direct video upload + API analysis
register-skill-release / register (push) Successful in 16s Details
- New detect-video / detect-video-and-search commands: upload video to get
  public URL, analyze via LiteLLM (video_url), generate keyword, search 1688
- New src/video-analyzer.ts: upload via direct HTTP (bypasses auth-rt CLI
  arg length limit), analyze via Chat Completions with video_url content
- Frame-based pipeline robustness: quality pre-filtering (skip black/blurry
  frames), bounding box normalization/validation, crop failure tolerance,
  Vision ranking fallback to sharpness-based selection
- Improve ranking prompt: force pick one frame, Chinese description
- Update docs to recommend detect-video-and-search as primary command

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-25 16:30:01 +08:00
ywkj 91a623751d skill: translate user-facing docs to Chinese, add detect-best commands
register-skill-release / register (push) Successful in 22s Details
- SKILL.md / README.md: full Chinese translation for Chinese users
- scripts/run.ts: help text in Chinese
- src/: add detectBest and detectBestAndSearch commands

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-25 15:13:07 +08:00
ywkj 33e3d378cc feat: update skill
register-skill-release / register (push) Successful in 17s Details
2026-04-22 08:23:35 +08:00
ywkj 778a51ad45 feat: use OpenClaw sub-agent for slow detect commands, fix keyword generation
register-skill-release / register (push) Failing after 1m36s Details
- SKILL.md: detect/detect-and-search now spawned via sessions_spawn (non-blocking); search/session run inline
- product-detector.ts: replace sequential chunk loop with worker-pool concurrency (withConcurrency) so all frames dispatch immediately up to the limit
- index.ts: fix generateChineseKeyword prompt to name the container/organizer object, not the items it holds (e.g. 鞋架 not 鞋)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-21 08:20:37 +08:00
ywkj 9f381f9bab refactor: remove VISION_API_KEY/BASE env overrides, all credentials from client config only
register-skill-release / register (push) Successful in 21s Details
2026-04-20 12:21:44 +08:00
ywkj 9d6a15f010 chore: set default vision model to aliyun-cp-multimodal 2026-04-20 12:18:01 +08:00
ywkj 23bb268d75 refactor: load vision config and search endpoints from auth client-config, no skill-level envs needed 2026-04-20 12:14:43 +08:00
ywkj fa03962216 docs: add missing env vars for image search and telemetry 2026-04-20 12:08:38 +08:00
ywkj 67abe94938 docs: add README 2026-04-20 12:06:20 +08:00
ywkj c3523d002e fix: 遥测改用 TELEMETRY_ENDPOINT,不复用 hookUrl
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-20 07:46:45 +08:00
ywkj fd018ccd6a feat: 接入 Hook 遥测回调
与 template-skill 保持一致,每次执行后自动 POST hookUrl 上报执行结果。

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-20 07:40:15 +08:00
ywkj 985e822ff2 chore: 接入 CI/CD skill 注册流程
register-skill-release / register (push) Successful in 29s Details
新增 .forgejo/workflows/register-skill-release.yml,推送 v* tag 时自动注册 skill。
install.sh 同步 template-skill,支持 auth-rt 自动安装。

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-20 07:26:42 +08:00
ywkj cabd1b332a feat: 初始化 video-product-snapshot skill
视频商品检测 + 1688 以图搜图 + 关键词二次过滤完整流程。

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-20 07:24:28 +08:00