orchestrate

SUSPICIOUS. The skill’s capabilities broadly match its orchestration purpose, and its visible toolchain/data flows point to local repo scripts plus standard Git/GitHub tooling rather than obvious credential theft or external exfiltration. The main risk is scope and autonomy: it can unattendedly read tracker content, launch subagents, execute commands, modify code, mutate tracker state, and integrate changes, while safety depends on unseen local Python scripts and prompts. Not confirmed malware, but a high-impact automation skill with meaningful execution risk.

The code appears to implement a Git/GitHub issue orchestration workflow and contains no clear malware indicators or credential theft/exfiltration behavior. The principal security concern is deliberate command execution from repository-controlled generated.json via shell=True, compounded by execution of adjacent model-selector scripts and extensive GitHub/Git mutations. Treat generated declarations and neighboring scripts as trusted code, validate or sandbox generator commands, avoid shell=True where possible, and review repository write permissions. The supplied fragment is syntactically invalid as shown, so execution requires correction or the source may be truncated.