ScreenLane v0.2.0

ScreenLane v0.2.0 adds the local Android workflow foundation: action planning, approval requirements, sensitive-app policy blocks, local receipts, and SDK/API support.\n\nPublished artifacts:\n- npm: @talocode/screenlane@0.2.1. This npm-only patch fixes the packaged screenlane CLI mapping.\n- PyPI: talocode-screenlane@0.2.0. This release adds Python Android workflow client support.\n\nThe package patch versions intentionally differ. PyPI 0.2.1 does not exist.

ScreenLane v0.1.4

ScreenLane v0.1.4

Talk to your screen.

Auth & cloud (simple)

Key	TALOCODE_API_KEY
Cloud API	https://api.talocode.site

Local capture / command / demo work without a key.

Install

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

pip install talocode-screenlane

What's new in 0.1.4

• Local OCR via tesseract for screen/image captures
• screenlane command --context-file image.png OCR + fix-error flow
• Cloud providers probe Talocode-documented routes
• Professional 48s demo video (talocode-video skill / motion graphics)
• Docs: only TALOCODE_API_KEY + api.talocode.site

Demo

screenlane demo
screenlane capture --source file --file ./error.png
screenlane command --text "Fix this error" --context-file ./error.png --target codra --out prompt

Cloud status (honest)

GET https://api.talocode.site/health is live (Stacklane API host).
Product chat routes (/v1/router/chat/completions, /v1/tera/*, /v1/codra/*) return 404 on the current deploy until the full Talocode Cloud router is live. Offline clipboard/stdout send always works.

Limitations

• Live mic not bundled — use --text
• OCR needs tesseract-ocr installed
• Screenshots need OS tools when available

Links

• https://github.com/talocode/screenlane
• https://www.npmjs.com/package/@talocode/screenlane
• https://pypi.org/project/talocode-screenlane/

ScreenLane v0.1.1

Fix npm bin entry so npx @talocode/screenlane works. Core product same as v0.1.0.

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