agentleFS
full-aigc-pluginsfull-aigc-plugins/dreamina-design-plugin/skills/dreamina-design-harness/SKILL.md

dreamina-design-harness

Skill0 starsChanged 23 days agoApache-2.0View on GitHub

Dreamina Design invocation spec for WorkBuddy - the image/video generation CLI (trusted_cli), text2image/image2image/text2video/image2video families across cli/opencli/prompt variants, shot annotation, video evaluation, and the paid submit-once gates with real submit_id verification. Read this before any Dreamina generation task.

The file

Files
SKILL.md

即梦设计调用规范(智能体通用)

执行通道:本插件随附的生成 CLI(scripts/trusted_cli.py 为受信入口;dreamina-cli 技能含完整用法)。 后端:即梦(Dreamina),付费动作,submit-once 门禁与真实 submit_id 验收是硬约束。

1. 能力族(24 个技能已随插件分发)

族技能
文生图 / 图生图dreamina-cli-text2image / image2image(+ opencli-* 变体 + prompt-* 提示词口径)
文生视频 / 图生视频dreamina-cli-text2video / image2video(同上三变体)
视频生产流dreamina-video-production(生产编排)、dreamina-shot-annotator(镜头标注)
评测dreamina-video-evaluator(视频质量评测)、dreamina-vision-judge(目标图 vs 候选图的独立评审)
视觉质量闭环dreamina_visual_loop MCP 工具:锁定目标 → 一轮付费生成 → 宿主独立评审 → 受额度约束的一次重试
目标获取dreamina-visual-target:用户没给目标图时取得目标(基线精修优先),反演绎纪律与付费门禁
长片续跑dreamina-auto-seedance(自动续跑)、dreamina-seedance-resume(断点恢复)
入口dreamina-design-use / dreamina-cli

边界(如实):设计套件专注图与视频,无独立音频生成技能;声音能力在"即梦画布"套件 (dreamina-canvas-generate-audio),需要配音/音效时转画布团队。

2. 付费门禁(不可绕)

  • 每次付费提交都要:用户明确授权 → 记录真实 submit_id → 凭据核对(回执/下载的 SHA-256 等)。
  • submit-once:一次授权一次提交;失败恢复走技能里的恢复路径,不得自动重试付费提交。
  • 预算上限由用户给定;报价缺失时(QUOTE_UNAVAILABLE)停下问,不猜。
  • 视觉质量闭环的重试另受精确请求指纹 + 信用额度上限约束:首轮评价未达标只进入等待授权, 不会自行提交第二次;授权绑定的指纹或额度与实际请求不一致时一律拒绝。
  • 目标生成同样是一次付费提交,须经用户授权;无授权时停下问,不静默降级为无目标迭代。

3. 标准工作流

  1. 明确产物类型(图/视频)与提示词基线(prompt-* 技能)。
  2. 走 cli 或 opencli 族执行生成(参数以技能文档为准)。
  3. 视频过 dreamina-video-evaluator;镜头类需求用 shot-annotator;图像对标迭代用 dreamina-vision-judge。
  4. 要迭代到与目标图一致时走 dreamina_visual_loop:用户没给目标图时先用 dreamina-visual-target 取得目标(目标生成是付费提交,须授权;已有产物走基线精修,不另起方向),lock_target 锁定 → run_first_round 执行一轮付费生成 → 把返回的证据交 dreamina-vision-judge 评审 → record_judgement 记录结论。一轮由两个调用组成(生成、记录评审),因为评审必须由宿主的全新上下文子代理完成 ——宿主侧的适配配方见 references/judge-port-adapter.md。
  5. 交付:产物路径、submit_id、验证凭据、消耗与剩余预算、未验证项。

交付回执至少保留真实提交标识和验证状态,例如:

{"submit_id":"<provider-submit-id>","artifact_verified":true,"paid_retry":false}

4. 纪律

  • CLI 输出与回执是事实来源;叙述与输出冲突以输出为准。
  • 修复实现委托给 worker 子代理时,遵守 references/repair-delegation.md: worker 全新空上下文、只给高层指令、只实现不验收,实现与评审角色分离。
  • 与图片工厂(Codex 后端)是双轨:用户没指定后端时,问清按哪个账本走。

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.