Back to .md Directory

CCWAI VS Code Extension Status

- **Extension workspace scaffolding** — Added dedicated `extension/` package with manifest, build/publish scripts, TypeScript config, and dependency wiring so the VS Code bundle compiles alongside `core`/`cli`.

May 2, 2026
0 downloads
0 views
ai prompt openai gemini copilot workflow automation
View source

CCWAI VS Code Extension Status

✅ Completed (v0.4.0 Phase 1 Foundation)

  • Extension workspace scaffolding — Added dedicated extension/ package with manifest, build/publish scripts, TypeScript config, and dependency wiring so the VS Code bundle compiles alongside core/cli.
  • Activation + commands — Implemented ccwai.analyzeContext, ccwai.planWithAI, ccwai.runCommand, ccwai.sendContext, and ccwai.showThoughts, each invoking the shared CCWAIEngine, streaming to the CCWAI output channel, and optionally persisting results.
  • Context extractor pipeline — Captures active editor, selection, workspace-relative path, and timestamp, trims content intelligently, and feeds formatted snapshots into both CCWAI and external providers.
  • Context panel webview — Sidebar shows summary/snippet, git status (git status --short --branch fallback to git diff --stat), and the three most recent extension thoughts, with buttons to trigger Analyze/Send actions via webview messaging.
  • Thoughts integration — Extension writes Markdown artifacts into thoughts/shared/extension, exposes a quick-pick viewer, and keeps the panel in sync after each run.
  • Provider bridge v1 — Detects installed AI extensions by command IDs (Copilot/Codex/Gemini/DeepSeek), sends prompts directly when possible, or falls back to browser launch + clipboard hand-off; user preference set via ccwai.preferredExternalProvider.
  • Provider routing indicators — Explicitly map known Copilot, Gemini, Codex, and DeepSeek command IDs, detect their availability, and surface their status inside the context panel so users know whether prompts go direct or via browser clipboard.
  • Engine bridge — Safely discovers the CCWAI project root, validates ccwai.config.json/commands, and instantiates CCWAIEngine with process cwd juggling so CLI + extension share logic.
  • Workspace buildnpm run build --workspaces now compiles core, CLI, and extension, ensuring a single command validates the full monorepo.

🚀 Next Steps (Prioritized TODOs)

  1. First-run / onboarding experience
    • Add welcome panel or notification guiding users through configuring providers, running Analyze Context, and locating saved thoughts.
  2. Thoughts explorer view
    • Replace the quick-pick list with a tree view showing shared vs personal thoughts; allow opening, deleting, or exporting directly from VS Code.
  3. Authentication / API key helpers
    • Provide commands/UI to set OpenAI/Gemini/DeepSeek keys or launch OAuth/login flows for users without local keys.
  4. Command palette macros
    • Add presets (e.g., “CCWAI: Analyze Current Git Diff”, “CCWAI: Draft PR Summary”) that prefill prompts using the context extractor + git summary.
  5. Release automation
    • Create CI workflow to lint, build the extension, run smoke tests, and produce a .vsix; document the publish process for the VS Code Marketplace.
  6. Telemetry & diagnostics
    • Optional (opt-in) logging of command usage/errors, surfaced via ccwai doctor inside VS Code to aid troubleshooting.
  7. Documentation refresh
    • Update README.md / docs with extension installation steps, screenshots, and the new settings (preferredExternalProvider, autoSaveThoughts, contextDepth).

We can now work through these in order—starting with provider routing polish and onboarding—while keeping the extension shippable at each increment.

Related Documents