Dangerous exec
Critical
- Finding
- Shell command execution detected (child_process).
Agent Swarm Dev Team
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill's high-level purpose (multi-agent developer orchestration) is plausible, but there are several internal inconsistencies and risky defaults (hardcoded user path, missing referenced scripts, and dangerous config flags) that merit review before running anything.
Do not run the init script without review. Specific steps to consider: 1) Inspect the missing scripts (run-agent.sh, check-agents.sh) before executing anything — they are referenced but not included. 2) Edit the hardcoded SKILL_DIR path to a relative or user-resolved location (avoid /Users/xufan65/...), or run in a disposable container/sandbox. 3) Remove or understand any flags like "--dangerously-bypass-approvals-and-sandbox" and "--dangerously-skip-permissions" in config — these are red flags and should not be used. 4) Decide how notifications and gh CLI auth will be provided (do not hardcode tokens); verify where data (PR status, logs) will be sent. 5) If you plan to use this, run it in an isolated environment (VM/container) first and provide the maintainers' missing runner scripts or request them before trusting the skill. If possible, request the author to explain the dangerous flags and to fix path and script inconsistencies. Additional information that would raise confidence: the contents of run-agent.sh and check-agents.sh, an explanation for the defaultArgs values, and a non-hardcoded installation path.
Static analysis
VirusTotal
No VirusTotal findings
Risk analysis
No visible risk-analysis findings were reported for this release.