MIT-licensed Amazon listing / A+ image workbench as a native dsh plugin — slot planning, English prompt assembly, one-image-at-a-time apiz generation on top of your own product photos, plus local synthetic-performer XMP tagging.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-11-20
License
MIT
01What can it help you accomplish?
Turn pasted Amazon listing copy into a per-slot image plan for MAIN + PT01–06 or one of four A+ module sets
A slot-by-slot plan JSON — Chinese rationale, English generation prompt and negative words per image — validated and archived under a planId
Amazon sellers and listing operators who need images that follow what each Amazon slot is supposed to prove
Generate the images one at a time, always anchored to your own product photos
A generated PNG per slot with the real reference image attached, plus actual size, quality tier, task_id and credit spend in the generation history
Sellers who must ship an image of the exact product they will ship, not a look-alike
Tag AI-generated performer media for Amazon compliance without uploading it anywhere
`contains-synthetic-performer` XMP written into JPG/PNG/WebP/MP4/MOV files locally
Anyone publishing synthetic-performer photos or video on Amazon
02How to install into DeepSeek Harness
Prerequisites
- dsh >= 0.2.1-rc.2 < 0.3.0 and Node 22+
- An apiz `sk-` key, or an existing `apiz auth login` session to reuse
- apiz CLI installed locally if you want reference images on the apiz channel (附图 / A+)
Installation steps
- 01
Add it from the panel: 侧栏「插件」→ 添加插件 → fill in an absolute local path or package name → enable the bundle after install
- 02
Or install from the CLI: `dsh plugin --profile web add https://github.com/liangdabiao/dsh-amazon-image-studio`
$ dsh plugin --profile web add https://github.com/liangdabiao/dsh-amazon-image-studio
- 03
Open the 「亚马逊图片」 sidebar entry and fill the apiz key on the 设置 page (or tick reuse of the local `apiz auth login` session) — generation is refused otherwise
Verify the integration
- The sidebar shows a 「亚马逊图片」 entry — the README's post-install check
- From a source checkout, `npm run selftest` checks bundled assets, the skill provider and engine critical paths
03DSH integration and capability boundaries
Native dsh plugin: registers the `amz_*` tools (`amz_slots`, `amz_plan`, `amz_validate_plan`, `amz_prompt`, `amz_generate`, `amz_tag_performer`, `amz_history`), ships the `amazon-image-studio` skill and mounts the 「亚马逊图片」 sidebar panel
Slot and spec lookup (amz_slots)
none — deterministic tables→listing slot table (MAIN + PT01–N), four A+ module layouts with official upload sizes and image-generation sizes
AI planning + validation gate (amz_plan, amz_validate_plan)
listing copy plus up to 8 MiB of reference images→strict-schema slide-by-slide plan archived as a planId, or a full self-planning brief for the current chat model
calls the configured planning model endpoint when a planning key is setFinal prompt assembly (amz_prompt)
a plan slot plus style board, density and target size→English prompt with style block, series style guide, negative words and a style-copy guard
skips the style board for MAIN, which must stay pure whiteOne-image-at-a-time generation (amz_generate)
approved prompt, product reference images, quality tier and resolution→PNG on disk plus history entry (generationSize vs actualSize, quality, task_id, prepaid credits)
uploads each reference image to the apiz CDN once to obtain a public URLpre-deducts credits at task creation and polls until completionSynthetic-performer XMP tagging (amz_tag_performer)
JPG/PNG/WebP/MP4/MOV files→files tagged with `contains-synthetic-performer`, already-tagged files skipped
04Who is it for? When not to use it?
Good for
- Amazon sellers and listing operators who need images that follow what each Amazon slot is supposed to prove
- Sellers who must ship an image of the exact product they will ship, not a look-alike
- Anyone publishing synthetic-performer photos or video on Amazon
Not for
- Generation is refused without apiz credentials: an `sk-` key on the 设置 page or reuse of the local `apiz auth login` session. Planning itself needs no extra key, because the current chat model does the planning.
- Billing is by actual token usage at 100 credits = ¥1, pre-deducted when a task is created; a polling timeout must not be re-submitted, because a new task charges again.
- On the apiz channel every reference image leaves the machine: apiz image-to-image only accepts public HTTP(S) URLs, so the plugin uploads product photos and style boards to the apiz CDN first.
- It targets dsh >= 0.2.1-rc.2 < 0.3.0 with Node 22+, and only apiz is a supported image channel — the volces / openai / aliyun adapters stay in the code but are hidden from the panel and out of the acceptance scope.
05Compatibility, maintenance and safety notes
- Generation is refused without apiz credentials: an `sk-` key on the 设置 page or reuse of the local `apiz auth login` session. Planning itself needs no extra key, because the current chat model does the planning.
- Billing is by actual token usage at 100 credits = ¥1, pre-deducted when a task is created; a polling timeout must not be re-submitted, because a new task charges again.
- On the apiz channel every reference image leaves the machine: apiz image-to-image only accepts public HTTP(S) URLs, so the plugin uploads product photos and style boards to the apiz CDN first.
- It targets dsh >= 0.2.1-rc.2 < 0.3.0 with Node 22+, and only apiz is a supported image channel — the volces / openai / aliyun adapters stay in the code but are hidden from the panel and out of the acceptance scope.
MIT · no tagged release yet; last push 2026-11-16 (created 2026-11-15, 22 stars)
06Frequently asked questions
Do I need an API key?
For generation, yes: paste an apiz `sk-` key on the 设置 page or enable reuse of the local `apiz auth login` session — with neither, both the panel and the tools refuse. Planning needs no extra key: 「✨ AI 策划」 just fills a new conversation with the listing copy, slots and field rules, and your current chat model plans.
Why does it refuse to generate without a product photo?
Because a generation model has never seen your product: without a reference it invents something that looks close but is not yours, which on Amazon means returns and bad reviews. Reference images are therefore a hard gate, and a style board does not count as one.
What if polling times out before the image appears?
The task may still be running, and credits were already pre-deducted at creation. Do not press generate again — take the `task_id` from the error and query `apiz tasks get <task_id>`; a genuine failure is refunded in full, while resubmitting really does charge twice.
Can I use a provider other than apiz?
Not from the panel. The volces / openai / aliyun adapters are still in the code and run if you hand-edit `imageProvider`, but they are explicitly outside this project's supported surface — the docs and acceptance runs do not cover them.
Where do plans, images and history go?
Under `${DSH_HOME}/amazon-image-studio/`: `plans/*.json` for archived plans, `images/` for generated PNGs, `references/` for the product photos you uploaded and `history.json` for task id, spend and actual size. `DSH_HOME` defaults to `~/.dsh`, and deleting those folders is the documented cleanup.
07Related DSH workflows
dsh-image-gen
by shanliuling
Adds in-chat multimodal image generation via a `generate_image` tool with multi-provider support (Gemini/OpenAI/Seedream), interactive preview, copy/download, and session-attached results.
dsh-vision-router
by ysr666
Eyes for text-only DeepSeek Harness agents: built-in free vision chain (no key) + pixel-level vision tools (Q&A, grounding, crop, pixel diff, colors, OCR, SVG trace, cutout, screenshots). One-command install, no Python, image turns work like ordinary tool-calling turns.
dsh-vision-toolkit
by anionex
为纯文本 DSH Agent 提供 10 个结构化视觉工具:意图感知图片问答、长截图 OCR、原始像素 grounding、UI 还原、像素 diff 等
dsh-see-now
by saos-3624
Captures screen, OBS, or webcam frames, filters by visual change, packs them into image sheets, and delivers them to DSH sessions.
08Data and sources
DeepSeek Harness(dsh)插件:**Amazon Listing / A+ 图片 AI 策划 + 逐张生图 + 合规打标**。
| 侧栏面板 | 策划台(双模式表单 + **产品参考图拖拽上传** + 风格预设色卡 + 逐槽位 Prompt 卡片:预览/复制/生成/填对话)、策划档案、生成历史、AI 人物打标(拖文件写入 + 下载)、工作流与合规指南 |
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-11-20. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.
