Skip to content

Running the App

Terminal window
cd cyberpaw
npm run tauri dev

The frontend hot-reloads on file changes. The Python sidecar must be rebuilt manually after changes to agent/.

On startup the terminal prints:

Working directory: ~/your/project
⠋ Loading gemma-4-E2B-it-Q4_K_M.gguf…
Model ready.

Terminal window
cd cyberpaw
npm run build && npm run tauri build

Output: src-tauri/target/release/bundle/macos/CyberPaw.app


Type your request at the prompt and press Enter.

KeyAction
EnterSubmit input
Ctrl-CInterrupt the running agent
/ Navigate input history

Prefix a command with ! to run it directly as a shell command, bypassing the LLM:

❯ ! ls -la

Click the icon in the top-right to open Settings:

  • Working Directory — the root directory the agent operates in
  • Models — download and load models; installed models show a ✓ Installed badge
  • Permission Mode — controls which tool calls require approval
  • Context Window — token budget (auto-calculated from RAM by default)
  • Network Access — enables WebFetch, WebSearch, and Playwright tools

Terminal window
# Unit tests
cd cyberpaw/agent
../.venv/bin/pytest test_task_tools.py -v
# Integration tests (requires a model at ~/CyberPaw/models/)
PYTHONPATH=. ../.venv/bin/python ../tests/test_integration_tool_call.py
PYTHONPATH=. ../.venv/bin/python ../tests/test_model_loading_flow.py