RapidDraft Studio¶
Operational source:
/Users/adeelyj/Library/CloudStorage/OneDrive-Personal/100_Knowledge/203_TextCAD/01_Product_Project_Management/00_Project_Management_n_skills/01_tracks/rapiddraft-studio
This page is the local wiki mirror of the operational MASTER_PLAN.md and TRACKING.md.
Last updated¶
Operational source last updated: 2026-06-03
Wiki mirror refreshed: 2026-06-18
Current overall score¶
RapidDraft Studio overall score: 81 / 100
Prototype-baseline posture remains stronger at roughly 92 / 100, but the fuller workstation
target is the better current lens. No score increase is assigned for CAD Drive because that work
lives in RapidDraft Core. The deployed web Agent foundation is real, but Home and Expert still
need convergence before it counts as product-grade Studio behavior.
Current posture¶
RapidDraft Studio remains the flagship AI-first engineering workstation and chat-first TextCAD
surface. The long-term shell is grounded in the Tauri Studio repo, while the immediate Studio
lesson comes from the RapidDraft Core .io Agent testing app: Agent must feel like one continuous
assistant across Home and Expert.
Milestone scorecard¶
| Milestone group | Score / 100 | Current state | Main remaining gap |
|---|---|---|---|
| Workspace foundation | 97 | Strong and validated | Mostly polish-only residual work. |
| Engineering workspace depth | 90 | Working and validated | More natural prompting and quieter diagnostics. |
| Drafting, planning, report outputs | 86 | Working and validated | Richer planning/report integration. |
| Vision and fusion review | 16 | Reference-only | No repo-local parity yet. |
| Review-station collaboration | 20 | Reference-only | Active in Core/CAD Drive, not Studio repo. |
| Candidate remediation loop | 14 | Reference-only | No real candidate-session workflow yet. |
| Drawing-analysis and DraftLint mode | 8 | Not started in this repo | Entire lane missing here. |
| Launcher and multi-mode workflow | 86 | Working and validated | Later stations need deeper behavior. |
| Product-grade convergence and hardening | 74 | Working, still converging | Shell convergence and warning cleanup. |
Feature scorecard¶
| Feature track | Score / 100 | Current note |
|---|---|---|
| Chat-first workspace UX | 90 | Strong direction, still needs utility-surface polish. |
| Artifact system and timeline orchestration | 96 | Mature for current prototype scope. |
| STEP and viewer workflow | 92 | Strong, with richer anchor semantics ahead. |
| CNC DFM workflow | 87 | Strong baseline with deeper rules ahead. |
| Approval and write safety | 92 | Core contract is solid. |
| Validation harness and evidence system | 95 | One of the strongest assets. |
| Launcher and modes | 82 | Real but not equally complete across stations. |
| Frontend convergence | 74 | Main active polish frontier. |
| Vision and fusion workflow | 15 | Outside current local implementation. |
| Collaboration workflow | 20 | Prove CAD Drive in Core before porting concepts. |
| Agent Home/Expert continuity | 56 | Foundation exists; UX still feels split. |
Active update¶
2026-06-03 - Agent seamless UX becomes the active Studio convergence lane¶
Claude, Cursor CLI, and Codex review converged on the same UX diagnosis: the Agent works, but Home and Expert feel disconnected. The current mental model should become "one agent, one conversation, two viewports."
The week-1 slice is shared composer, honest Ask/Plan modes, Home session continuation, smoother Home-to-Expert handoff, and a calmer Expert panel hierarchy.
Highest-value next actions¶
- Run the Agent seamless UX week-1 slice.
- Keep broader Studio shell work parked while CAD Drive is shared and hardened in Core.
- Continue Milestone 22 only if it does not interrupt collaboration or Agent convergence.
- When parked milestones return, restart with vision/fusion before deeper collaboration or remediation.
Sources¶
- Operational Studio master plan and tracker.
01_tracks/rapiddraft-studio/plans/260603_rapiddraft-agent-seamless-ux/README.md