Satellite Copilot

Predict satellite passes (NOAA APT, METEOR LRPT, ISS) for a configured latitude/longitude and send WhatsApp alerts with manual dish alignment info (AOS/LOS azimuth+elevation, track direction, inclination). Use when setting up or operating a zero-AI pass scheduler/orchestrator for SDR satellite reception, including configuring NORAD IDs, minimum elevation, alert lead time, and optional remote capture/decode hooks (Pi RTL-SDR capture + Jetson SatDump decode).

This repository appears to do what it claims, but please review and take these precautions before enabling automation or capture: 1) Keep capture/ decode entries disabled until you’ve inspected and tested the exact commands you will run — they execute in a shell (shell=True) and can run arbitrary code on your host or remote host. 2) Don’t rely on the code’s auto-accept SSH behaviour; manually SSH to your Pi/Jetson once from this user to verify host keys or modify the code/command to avoid StrictHostKeyChecking=accept-new. 3) Ensure the 'clawdbot' CLI (used to send WhatsApp messages) is the genuine CLI you expect and is not a malicious binary earlier in your PATH. 4) Be aware the predictor fetches TLEs from a public third-party service (tle.ivanstanojevic.me) — if you need offline operation, replace/fork that fetch or vendor TLEs. 5) Secure the config file (chmod 600 is suggested in README) and limit who can edit it; review any cron entries before installation. If you want a lower-risk test: run the scripts with config.enabled=false and step through code locally, or run them in an isolated container/VM before using them on a production host.

Type: OpenClaw Skill Name: moltbot-satellite-copilot Version: 0.1.0 The skill bundle is designed for satellite pass prediction and WhatsApp alerts, with optional remote capture/decode hooks explicitly marked as 'TODO' and disabled in the primary `scripts/orchestrator.py` script. While `orchestrator.py` defines an `ssh` function (a high-risk capability) and uses `StrictHostKeyChecking=accept-new`, it does not currently execute user-defined commands via SSH. The documentation (`SKILL.md`, `README.md`) is transparent, clearly stating the 'zero-AI' nature and that remote execution is 'in planning and development'. Another script, `scripts/radio_scheduler.py`, contains a `subprocess.run` call with `shell=True` for user-defined commands, which is a significant risk, but this script is not the one instructed for execution by the skill bundle's documentation.