Releases: truman-t3/keyshot-mcp
Releases · truman-t3/keyshot-mcp
Release list
v0.4.1
v0.4.0
English
New features
keyshot_list_cameras— camera discovery: open a scene and get the list of camera names, ready to feed into batch rendering.- Material preset library — reuse named materials instead of exact KeyShot names:
keyshot_list_material_presets— list presets frompresets/materials.json(override withKEYSHOT_MATERIAL_PRESETS).keyshot_apply_material_preset— apply a named preset to an object.
keyshot_render_queue— run several render jobs sequentially; stops at the first failure unlesscontinueOnErroris set, and reports per-job status (succeeded / failed / skipped).
Internals & tests
- New modules
src/presets.tsandsrc/queue.ts(the queue runner is injectable so it can be unit tested without KeyShot). - Test suite grew to 49 tests (41 Vitest + 8 Python); type check clean;
dist/rebuilt.
Version 0.3.0 → 0.4.0. Commit: 606f1c4
中文说明
新增功能
keyshot_list_cameras—— 自动发现相机:打开场景后返回所有相机名称,可直接用于批量渲染。- 材质预设库 —— 用"好记的名字"复用材质,不必记 KeyShot 精确材质名:
keyshot_list_material_presets—— 列出presets/materials.json里的预设(可用KEYSHOT_MATERIAL_PRESETS覆盖路径)。keyshot_apply_material_preset—— 把某个命名预设应用到物体上。
keyshot_render_queue—— 顺序执行多个渲染任务;默认遇到第一个失败即停止,设置continueOnError可跑完全部,并逐个返回状态(成功 / 失败 / 跳过)。
内部实现与测试
- 新增模块
src/presets.ts、src/queue.ts(队列执行器可注入,便于在没有 KeyShot 的环境下做单元测试)。 - 测试增至 49 个(41 个 Vitest + 8 个 Python);类型检查通过;
dist/已重新构建。
版本 0.3.0 → 0.4.0。对应 commit:606f1c4
v0.3.0
English
Highlights
- Fix
import_model: now actually opens thebaseScenePathbase scene before importing the model (previously the base scene was ignored). - Fix
set_camera: no longer crashes when the camera object cannot be obtained — it falls back to thelux-level API automatically. - Temp file cleanup: the runner now deletes its temporary
args/resultJSON after every run, sowork/tmpno longer grows unbounded. - Clearer
call_variantssemantics: the first callback that does not throw wins (returningNonecounts as success); only when all fail does it raise. - Test suites added: 25 Vitest (TypeScript) + 5 Python tests covering the fixes above as regression guards.
- Version bumped
0.2.1→0.3.0.
Commit: 3009564
中文说明
主要更新
- 修复
import_model:现在会真正打开baseScenePath指定的基础场景,再导入模型(之前基础场景被忽略了)。 - 修复
set_camera:拿不到相机对象时不再崩溃,会自动回退到lux级 API。 - 临时文件清理:每次运行后自动删除临时的
args/resultJSON,work/tmp不再无限膨胀。 - 澄清
call_variants语义:第一个不抛异常的调用即生效(返回None也算成功),全部失败才报错。 - 新增测试套件:25 个 Vitest(TypeScript)+ 5 个 Python 测试,为上述修复提供回归保护。
- 版本
0.2.1→0.3.0。
对应 commit:3009564
v0.2.1 - LobeHub Metadata Improvements
- Added official LobeHub MCP badge for marketplace ownership verification.
- Added one MCP prompt: keyshot_product_render.
- Added one MCP resource: keyshot://workflow.
- Added prepare script for GitHub-based installs.
v0.2.0 - Batch Render Support
Second public release of KeyShot MCP.
Highlights:
- Added
keyshot_batch_renderfor rendering multiple named cameras from one scene - Added npm metadata: repository, homepage, bugs, and Node.js engine requirement
- Clarified Windows + KeyShot Studio + Codex positioning
- Updated README tool list and roadmap to reflect batch rendering support
Batch render parameters:
scenePathoutputDircameraswidthheightformatoverwrite
Validation:
- TypeScript build passed
- Python bridge syntax check passed
- MCP tool discovery passed with
keyshot_batch_render - KeyShot headless status check passed on Windows 11 / KeyShot Studio 2025 / 14.1
v0.1.0 - Initial KeyShot MCP Server
Initial public release of KeyShot MCP.
Features:
- Check KeyShot headless startup status
- Inspect KeyShot scene files
- Render scenes to image files
- Import model files
- Apply materials to objects
- Create or update cameras
- Set environment when supported by installed KeyShot version
- Save scenes to new files
Tested on:
- Windows 11
- KeyShot Studio 2025 / 14.1
- Node.js 22