Yang-Agent_Revit
5788324/Yang-Agent_Revit
面向 Revit 使用者、BIM 团队、插件开发者和公司内部自动化团队。 本文档目标是从 0 开始,规划一套基于 Codex、Claude Code、Antigravity、MCP、pyRevit、Revit API 的公司级 Revit AI Agent 工作流。它的目标不是让 AI “一键解决所有 Revit 问题”,而是把 AI 变成一个可靠的 Revit 开发和生产助手:能理解你的模型、理解你的操作习惯、帮你写代码、开发插件、检查问题、批量处理图纸和数据,并在关键操作前保留人工确认。
Instalação
npx convez-ai 5788324/Yang-Agent_Revit1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
YangAgent Revit
YangAgent Revit is a personal Revit AI assistant project.
The goal is simple: help YANG inspect Revit models, generate AI-readable reports, understand problems, and apply only low-risk fixes after human confirmation.
This is not a company platform, commercial plugin, or enterprise deployment system.
Current Authority Order
When documents conflict, use this order:
docs/product-brief.mddocs/project-rules.mddocs/framework/daily-ops-routine.mddocs/handoff-new-chat-2026-06-07.mddocs/next-steps.md- latest relevant worklog under
docs/worklogs/
Older architecture, MCP, Bridge, company, and commercial planning documents are historical reference only unless a current authority document points to them.
Current Focus
This week focuses on one result:
pyRevit MVP usable in a sandbox model
That means:
- the pyRevit toolbox loads;
- report and preview buttons run in a sandbox model;
- exported JSON/CSV/Markdown can be read by AI;
- model-changing tools use
preview/dry-run -> human confirmation -> apply -> log -> Undo check; - the first real sandbox blocker is captured and fixed.
Current Tools
System SettingsReport Export PathExport Model SnapshotModel Health ReportExport Regression ChecklistExport AI Review PromptPreview Missing MarksPreview Missing Room NumbersPreview Duplicate Room NumbersPreview Unplaced ViewsPreview View Naming RulesApply Missing Door Window MarksApply Missing Room Numbers
The toolbox supports Chinese and English.
Core Documents
- Product brief
- Simple roadmap
- Project rules
- Agent operating model
- Agent development rules
- Agent task template
- Agent delivery report template
- Agent review checklist
- Daily ops routine
- Document governance audit
- External toolbox intake
- Sandbox checklist
- Sandbox feedback template
- Current handoff
- New chat startup
- Next steps
Install pyRevit Toolbox
.\scripts\install-pyrevit-extension.ps1
If Revit loads stale buttons or cache errors, close Revit first, then run:
.\scripts\install-pyrevit-extension.ps1 -Force -ClearCache
Offline Preflight
Run before a live sandbox session:
python tools\check_pyrevit_extension.py
python tools\run_sandbox_preflight.py --write-report
python tools\static_checks.py --write-report
Expected current status:
- pyRevit preflight:
0 errors - sandbox preflight:
7/7 PASS - static checks:
0 errors, known historical warnings only
Longer-Term Direction
Long-term goals are still allowed, but they are not this week's delivery target:
- Revit 2022-2027 support;
- Gemini C# toolbox intake and selected migration;
- MCP automatic model reading;
- MCP controlled model modification through preview/apply only.
The project should grow only from real work needs.