agentleFS
full-aigc-skillsfull-aigc-skills/dreamina-skills/skills/dreamina-3d-from-blender/SKILL.md

dreamina-3d-from-blender

Skill4 starsChanged 59 days agoApache-2.0View on GitHub

Use when the user has a Blender scene and wants a Seedance 2.5 render from it. Drives the preview through the validated Dreamina 3D pipeline.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 3 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When the user provides one auto_with_budget envelope — approved reference upload, model/resolution/duration preferences, a maximum charge, …

    line 23
  2. Distinctive:

    Stop and ask only when the quote exceeds the cap, a required reference was not authorized for upload, a platform-mandated confirmation or …

    line 31
  3. Very specific:

    Inspect: Call inspect_scene(executable=blender_adapter, scene=<user_scene>).

    line 42
  4. Very specific:

    Export: Call request_preview_export(...) with the user's prompt metadata embedded in the request payload.

    line 46
  5. Very specific:

    Validate: Run validate_artifact(receipt, current_file) from scripts/handoff_validator.py.

    line 50
  6. Very specific:

    Resolve readiness: Use McpDesignClient to call dreamina_cli_status and dreamina_account.

    line 56
  7. Very specific:

    Never resubmit a job whose design_submit_id is already set.

    line 80
  8. Distinctive:

    Never skip the preview validation gate.

    line 81
  9. Distinctive:

    Never send DCC scene data to the design plugin — only the validated artifact reference (artifact_id + sha256), the user prompt, and the …

    line 82
  10. Very specific:

    Never use the consumer plugin's internal design_handoff.py fixture for a production route.

    line 85

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

dreamina-3d-from-blender

When to use

The user wants a Dreamina 3D render from a Blender scene. Companion blender-design is already installed.

Workflow

Automatic run policy

When the user provides one auto_with_budget envelope — approved reference upload, model/resolution/duration preferences, a maximum charge, and permission for one remote submission — perform the following workflow without asking at every state transition. Resolve capability, validate the preview, obtain and compare the quote, submit once only when it is within the cap, query the recorded identifier, download, and independently verify the final artifact. Return one final artifact inventory.

Stop and ask only when the quote exceeds the cap, a required reference was not authorized for upload, a platform-mandated confirmation or web prerequisite is pending, validation fails, or recovery would require another paid submission. interactive retains the review at each gate; review_only does not upload or submit.

Because Dreamina Design currently exposes no authoritative quote tool, auto_with_budget stops at QUOTE_UNAVAILABLE. The user may instead approve one fully specified auto_exact_request; that grants one submission only and does not bypass the native paid-action approval in Dreamina Design.

  1. Inspect. Call inspect_scene(executable=blender_adapter, scene=<user_scene>). Reject if the scene cannot be inspected.
  2. Specify preview. Collect user-approved camera, frame range, output path, and dimensions. Build a PreviewSpec.
  3. Export. Call request_preview_export(...) with the user's prompt metadata embedded in the request payload. Independent re-hash is performed by the orchestrator — never trust the producer's declared sha256 alone.
  4. Validate. Run validate_artifact(receipt, current_file) from scripts/handoff_validator.py. On any error, transition the job to Failed with the classified error_category. 调用方若需限定生产者版本,可传 version_ranges={"blender-design": [("1.0.0", "1.9.9")]}; 显式映射替换默认范围,{} 拒绝全部生产者,None 保留默认范围。 版本通过仍必须完成路径、字节数、摘要与其余元数据校验。
  5. Resolve readiness. Use McpDesignClient to call dreamina_cli_status and dreamina_account. On user action required, stop with the returned remediation.
  6. Bind request. Build QuoteInputs(prompt, model, resolution, ratio, duration) and the validated preview reference. Record that an authoritative quote is unavailable; never synthesize one.
  7. Authorize. Require auto_exact_request (or a future authoritative quote within auto_with_budget) plus reference-upload and one-submit flags.
  8. Submit. Let McpDesignClient call dreamina_submit_video; Dreamina Design obtains native approval internally. Persist submit_id before reporting Submitted.
  9. Query. Let McpDesignClient call dreamina_query_task for the stored identifier and approved download root. Unknown is query-only.
  10. Download + verify. Re-hash the downloaded artifact and compare to the declared hash; reject on mismatch.

Ledger transitions

Drive the job through Draft → DccSelected → PreviewSpecified → PreviewValidated → CapabilityResolved → Quoted → Approved → Submitted → Querying → Completed | Failed | Unknown exactly once each.

Never do

  • Never resubmit a job whose design_submit_id is already set.
  • Never skip the preview validation gate.
  • Never send DCC scene data to the design plugin — only the validated artifact reference (artifact_id + sha256), the user prompt, and the quote inputs.
  • Never use the consumer plugin's internal design_handoff.py fixture for a production route. It is not shipped by this granular skill and exists only as a deterministic compatibility harness in the plugin test suite.
<!-- QUALITY_BASELINE_V1 -->

When to use(什么时候使用)

当用户需要 为当前请求选择并执行可验证、可恢复的专业工作流 时加载本技能。先从请求中提取目标、输入、约束、交付格式和验收标准;描述摘要为:Use when the user has a Blender scene and wants a Seedance 2.5 render from it. Drives the preview through the validated Dreamina 3D pipeline.。

Rules

  • 先读后写:先确认当前状态与真实能力,再执行会改变外部状态的动作。
  • 权限最小化:只使用完成当前步骤所需的文件、工具、账户与网络范围。
  • 证据优先:运行结果、资源 ID、版本、哈希或测试输出缺失时,明确标记为 NOT_VERIFIED。
  • 幂等优先:保留请求标识与阶段状态;结果不明确时先查询,不进行盲目重试。
  • 隐私安全:日志、示例、回执和错误信息不得包含 token、cookie、密钥或个人敏感数据。

Workflow

Step 1:澄清意图

确认本技能是否匹配目标;若只是相邻需求,交给更精确的技能。

Step 2:执行预检

确认目标、输入、约束、可用工具、成功标准和失败边界;任一关键条件未知时停止在只读阶段。

Step 3:形成计划

列出将调用的工具、会改变的对象、成功标准以及失败后的安全退出方式。

Step 4:执行动作

按最小充分步骤执行,并在关键状态变化处记录证据;每个外部调用均保留可关联的状态或回执。

Step 5:验证交付

输出结果、验证证据、未完成项、风险和明确的下一步,并把事实、推断和未验证项分开陈述。

Validation checklist

  • 技能触发条件与用户意图一致,没有把相邻任务误路由到本技能。
  • 输入、目标对象、版本和输出位置均已明确,且没有使用猜测值替代必填值。
  • 所有写入、付费、发布或不可逆动作都在用户授权范围内。
  • 结果已用独立检查验证;仅有“命令成功”或“文件存在”不算完整验收。
  • 输出包含实际证据、失败/跳过项、剩余风险和可执行的下一步。

Gotchas

  1. 把计划当结果:文档或提示词不等于真实执行;必须标明实际运行层级。
  2. 错误重试:超时或响应丢失可能已经产生远端状态,先查询再决定是否重试。
  3. 隐式扩大范围:批量、全量、发布、覆盖和付费不是普通读写的自然延伸。
  4. 版本漂移:引用外部资源时记录版本、tag 或提交;不要把可变分支当发布证据。
  5. 证据过期:缓存、旧截图和历史测试不能证明当前环境;在交付前刷新关键证据。

不适用与边界

不超出用户给定范围;写入、付费、发布和不可逆动作需要明确授权。 如果请求需要别的技能,不复制其正文;按技能名称进行交接,并保留当前任务上下文。

Progressive disclosure

  • 需要确定输入/输出、状态和授权点时,读取 references/workflow-contract.md。
  • 需要交付前自检时,读取 references/validation-checklist.md。
  • 遇到超时、部分成功或恢复场景时,读取 references/error-recovery.md。
  • 首次运行、拒绝越权和失败恢复分别参考 examples/happy-path.md、examples/boundary-refusal.md、examples/failure-recovery.md。

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.