ScreenLane v0.1.0

ScreenLane v0.1.0

Talk to your screen.

Open-source screen-aware voice command layer for AI agents — capture screen context, transcribe voice, generate agent-ready commands, and send actions to Tera, Codra, Codex, OpenCode, GateLane, or MCP tools.

What's included

• CLI screenlane: init, capture, dictate, command, send, serve, mcp, doctor, demo, auth
• TypeScript SDK @talocode/screenlane
• Local HTTP API on port 3070
• MCP server (stdio) with 8 tools
• Talocode Skills: screen-aware-command, screen-aware-debugging, screen-aware-writing
• Python package talocode-screenlane / CLI screenlane-py (HTTP client + helpers)

Install

npm install -g @talocode/screenlane
# or
npx @talocode/screenlane@latest demo

Python:

pip install talocode-screenlane
screenlane-py --help

Demo

screenlane demo
screenlane command --text "Fix this error" --context-text "TypeError: ..." --target codra --out prompt
screenlane doctor

Auth

Local-first by default. To gate the HTTP API:

export SCREENLANE_REQUIRE_AUTH=true
export TALOCODE_API_KEY=your_key
# Authorization: Bearer <TALOCODE_API_KEY>

Only TALOCODE_API_KEY is used for access control (no separate SCREENLANE_API_KEY).

Limitations (honest)

• Live microphone capture is not bundled; use --text or --audio + OpenAI/Tera keys
• Screenshot capture depends on OS tools (grim, scrot, screencapture, etc.) with text fallbacks
• Tera/Codra/GateLane sends require live base URLs + keys — not faked
• Command generation in v0.1 is deterministic template-based (works offline)

Roadmap

• Native mic recording helpers
• OCR on screenshots
• Deeper local agent launchers (Codex/OpenCode)
• Richer cloud transcription providers

License

MIT