Documentation Strategy & Plan (dIKta.me V2)

Goal: Create comprehensive, dual-audience documentation for dIKta.me V2 (C# / WinUI 3), leveraging the foundation of V1. The documentation must accurately reflect the existing application architecture, user interfaces, and development patterns.

1. Documentation Scope

A. User Documentation (docs/user/)

Focuses on how to use the dIKta.me desktop application, configuring settings, and troubleshooting.

• getting-started.md: The critical "first contact" guide for non-technical users.
  • Installation and navigating the initial onboarding Wizard.
  • Educational explanations of core concepts (e.g., "What is STT (Speech-to-Text)?" and "What is an LLM (Large Language Model)?").
  • Setting up the first STT/LLM provider without overwhelming the user.
  • Interacting with the System Tray icon and background operation.
• features/dictation.md: Core dictation workflow, the main Control Panel interface, using global hotkeys, streaming vs. batch dictation.
• features/refine.md: The post-processing pipeline (Refine), configuring rewriting styles, text expansion, and system prompts.
• features/ask.md: Using the "Ask" pipeline for voice-driven prompt execution and in-place substitution.
• features/translate.md: Using the "Translate" pipeline for real-time translation of dictated audio.
• features/note.md: Using the "Note" pipeline for pure STT transcribing (bypassing LLM processing).
• features/quick-chat.md: Using the Quick Chat overlay window ("Chat" pipeline) for ad-hoc assistant interactions and context injection.
• features/oops.md: Using the "Oops" (Ctrl+Alt+V) hotkey to undo or fix mistakes on the fly.
• settings/: Comprehensive guide to all 12 Control Panel settings tabs:
  • general.md: Language, autostart, and basic app behavior.
  • account.md: Wallet balances, trial status, and dIKta.me authentication.
  • api-keys.md: Configuring BYOK (Bring Your Own Key) for Anthropic, Deepgram, Gemini, and OpenAI.
  • ai-engine.md: Global model selections and prompt assignments.
  • audio.md: Input devices, recording times, and audio ducking features.
  • dictation-modes.md: Managing CRUD operations for custom dictation presets.
  • modes.md: Configuring utility pipelines (Ask, Refine, Translate).
  • hotkeys.md: Mapping global keybindings.
  • Macros.md: Managing local text expansion and auto-corrections.
  • ollama.md: Local inference settings (Models, Keep-alive, Context Windows).
  • privacy.md: History retention, PII scrubbing (Levels 0-3), and logging.
  • control-panel.md: HUD visibility toggles.
• troubleshooting.md: Common Windows-specific issues (Scaling, Permissions, Audio Driver locking, DPAPI issues).

B. Developer Documentation (docs/dev/)

Focuses on how to build, extend, and maintain the dIKta.me V2 codebase.

• setup.md: Dev environment setup (VS 2022, .editorconfig, running the WinUI 3 project).
• architecture/: Expansions of the high-level ARCHITECTURE.md.
  • audio-pipeline.md: The NAudio implementation, input capture, Voice Activity Detection (VAD).
  • ui-mvvm.md: WinUI 3 structure, CommunityToolkit.Mvvm usage, Dependency Injection.
• api/: Core internal extension points.
  • stt-providers.md: Implementing ISTTProvider and IStreamingSTTProvider (covering existing implementations like Deepgram, GeminiAudio, Whisper).
  • llm-providers.md: Implementing ILLMProvider (Anthropic, Gemini, Ollama, OpenAICompat).
  • pipeline.md: The text transformation pipeline orchestration.
• testing.md: Running tests (xUnit), writing mocks, UI testing notes.

2. Task Logs & Phased Execution

3. Acceptance Criteria

Before any documentation phase is marked complete, the following criteria must be met:

1. Code Accuracy: Developer documentation must perfectly align with the main branch. e.g., STT docs must list exactly Deepgram, GeminiAudio, and Whisper.
2. UI Reality: User documentation screenshots and descriptions must match the WinUI 3 components (e.g., WizardWindow, ControlPanelPage, QuickChatWindow, SettingsWindow).
3. Audience Separation: Developer concepts (DI, MVVM, interfaces) must not leak into the User Documentation. User documentation must remain accessible to non-technical users.
4. Formatting: All files must be pure Markdown (.md), use active voice, and avoid platform-specific Markdown extensions unless natively supported by GitHub.
5. Navigation: All pages must be accessible from their respective section indices (user/index.md or dev/index.md).

4. Tools & Standards

• Format: Markdown (.md).
• Diagrams: Mermaid.js for flows and architecture.
• Images: Stored in docs/assets/ (To be captured later).