Provider, context, and Command safety roadmap (#33–#45)

[triggeredcode]

Summary

Completes Lucid's provider, context, Command-safety, and clarification-UI roadmap on the isolated feat/provider-context-roadmap branch.

Closes #33
Closes #34
Closes #35
Closes #36
Closes #37
Closes #38
Closes #39
Closes #40
Closes #41
Closes #42
Closes #43
Closes #44
Closes #45

Included

• provider-neutral AI client and measurement baseline
• secure Electron main-process execution and credentials
• LM Studio, Ollama, OpenAI, Anthropic, OpenRouter, Ollama Cloud, and custom compatible profiles
• rebuildable document context index and targeted prompt context
• checked structured Command execution across provider capability modes
• stable clarification cards for long questions and four long options
• conservative exact Command fast paths with normal patch, history, undo, and confirmation safeguards
• eval reporting that separates deterministic execution from model execution

Verification

• full CI: 1,518 passed, 22 skipped
• Electron production build: passed
• Command desktop E2E: 8/8 passed
• the corrected Fix Command Mode clarification card layout for long questions and multiple options #33 narrow-layout regression passes
• the only remaining reproducible full-desktop-suite failure is the pre-existing unrelated top-bar height assertion in page-look.spec.ts
• Docdex import diagnostics: clean

Notes

• No release tag is created here.
• The local planning document and local Docdex ignore entries remain uncommitted.