recamera-intellisense

# reCamera Intellisense

## Requirements

- `python3` (no external packages)
- Reachable reCamera HTTP API (default port `80`)
- Credentials stored in `~/.recamera/devices.json` (created automatically; declared in skill metadata)

## Security considerations

- **Credential storage**: Device tokens are stored in `~/.recamera/devices.json`. Protect this file with appropriate permissions (`chmod 600`) and do not place unrelated secrets there.
- **Plain HTTP transport**: Communication to devices uses HTTP (port 80) by default — data including images and tokens travels unencrypted. Configure HTTPS on your devices if operating on untrusted networks.
- **Trusted networks only**: The skill polls devices and downloads snapshot/image files. Only use it with cameras on networks you trust.
- **Camera-specific tokens**: Use dedicated per-camera tokens (`sk_xxx`). Do not reuse tokens shared with cloud services.
- **Source review**: The bundle includes full Python sources under `scripts/`. Review them to verify behavior matches your expectations before granting autonomous execution.

## Scripts

All scripts live under `{baseDir}/scripts` and accept **one JSON object** as CLI argument (optional for `detect_local_device` and `list_devices`).

- **`device_manager.py`**: add/update/remove/list/get device credentials, file download
- **`detection_manager.py`**: models, schedule, rules, events, event-image fetch
- **`capture_manager.py`**: capture status/start/stop, one-shot image capture

**Full API signatures and CLI schemas**: See [REFERENCE.md](REFERENCE.md)

## Agent rules

1. Always pass complete JSON; never use interactive prompts.
2. Use exactly one of `device_name` (preferred) or inline `device`.
3. Auth token format: `sk_xxx` (from Web Console → Device Info → Connection Settings → HTTP/HTTPS Settings).
4. To detect by label name: call `get_detection_models_info`, map name → label index, use index in `label_filter`.
5. Poll `get_detection_events` every 1–10s; pass `start_unix_m...