blender
已验证 · 实测可装 CheshireJCat
功能简介
DSH插件:完整Blender 3D建模、重建、渲染、验证、导出工作流
可用 — 实测通过,社区增长中
DSH插件:完整Blender 3D建模、重建、渲染、验证、导出工作流 实测能干净安装、正常启动。社区在增长,是个稳妥选择。
「已验证」表示我们的自动化 CI 在干净 profile 里实际执行了 dsh plugin add 并启动成功——仅此而已。功能描述与版本兼容性均为作者声明。这不是安全审计,也不代表对第三方代码的背书。
README
dsh-blender
An installable DeepSeek Harness plugin that gives dsh a complete Blender-backed 3D production workflow. It registers one orchestrator plus 29 directly loadable domain skills, all 26 upstream analysis/validation helpers, and 13 workspace-scoped Blender tools.
Example output

Camera-free preview generated during a real dsh-blender installation smoke test.
Tools
blender_statuschecks Blender, analysis Python, skills, helpers, and the active workspace.blender_scene_infoandblender_object_infoinspect scenes and object-level geometry, materials, UVs, constraints, shape keys, and animation.blender_importconverts supported portable assets into versioned.blendsources.blender_pythonruns a small reviewedbpy/bmeshchunk and saves a versioned.blend.blender_preview,blender_render, andblender_render_framescreate blockout, look-dev, and animation evidence forread_imageQA.blender_exportexports GLB/glTF, FBX, OBJ, STL, USD, or PLY.blender_validate_sceneandblender_validate_exportperform target-aware audits and clean-process re-import checks.blender_helper_catalogandblender_helper_runexpose all 26 deterministic reference, wireframe, contour, multiview, UV, texture, look, repair, and animation-QA helpers.- The bundled 30-skill stack coordinates blockout, refinement, materials, lighting, cameras, animation, reconstruction, validation, and artifact handoff.
No Blender add-on or control port is required: each operation launches a local Blender background process.
Requirements
- DeepSeek Harness
0.1.0-rc.6 - Node.js 20+
- Blender 4.3+ available as
blenderonPATH, or configured with an absolute path - Python 3.10+ for the optional reference-analysis helpers
Install from npm
npx @deepseek-ai/dsh plugin --profile web add dsh-blender
npx @deepseek-ai/dsh plugin --profile web exec dsh-blender-setup
npx @deepseek-ai/dsh --profile web --dump-config
npx @deepseek-ai/dsh web
The setup command creates a private .venv inside the installed package and installs OpenCV, NumPy, Pillow, and SciPy. Skip it only if you do not need reference, wireframe, multiview, texture, or animation-analysis helpers.
To install a pinned GitHub release instead:
npx @deepseek-ai/dsh plugin --profile web add github:CheshireJCat/blender#v0.2.2
Develop from source
pnpm install
pnpm setup:analysis
npx @deepseek-ai/dsh plugin --profile web add .
Open http://127.0.0.1:3080, create a session whose workspace is the modeling directory, and ask dsh to use create-3d-model.
See README.zh-CN.md for configuration, security boundaries, headless use, and validation commands.
安装
装一次目录插件,之后本站所有插件都能让 DeepSeek Harness 自动找、自动装:
dsh plugin add dshbase-catalog 然后对 agent 说「帮我装 blender」,它会在目录里找到并自动安装。文档:dshbase-catalog · 已验证场景包。
Web profile:
dsh plugin --profile web add blender Headless(CLI)profile:
dsh plugin --profile headless add blender 包信息
npm:blender · 版本 — · 实测环境 dsh 0.1.0-rc.6
实测报告
端到端验证通过:dsh 0.1.0-rc.6 上 L1 安装 + L2 加载 + L3 运行问答。
使用场景
给 agent 一套记忆、知识库或检索层,让它不再跨会话丢上下文。
适合谁
跑长项目、想让 agent 记住决策、文档和偏好而不用每次重讲的人。
二次开发建议
记忆/检索后端是缝——插新存储、调蒸馏策略,或加引用与审计轨迹。