Agent Swarm Network

• Analyzed: May 15, 2026, 4:56 AM
• Findings: 1
• Version: 1.0.6
• Hash: 2cc2987d5783354431935cdb12113d1744d8cff33b99b363606133f6af8aaa12
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill is classified as suspicious primarily due to the explicit disclosure of a critical data-at-rest vulnerability: context snapshots, which may contain sensitive PII and API keys, are stored unencrypted in `~/.pilot/inbox/` (SKILL.md, README.md, manifest.json). While the skill transparently warns users and advises `chmod 700` for mitigation, this inherent risk of sensitive data exposure warrants a 'suspicious' classification. All operations are confined to the `~/.pilot/` directory and network communication defaults to localhost, but the capability to configure remote network endpoints exists, which could be exploited if the agent were maliciously prompted. There is no evidence of intentional malicious behavior such as exfiltration to unauthorized external endpoints or stealthy persistence mechanisms by the skill itself.