auditing-security

Security auditing identifies vulnerabilities in Skill code by scanning SKILL.md, Python, Shell, and JavaScript scripts for hardcoded credentials, network calls, obfuscated code, and data leakage risks. The audit reviews file-by-file patterns, access controls, and configuration files to detect suspicious activity and misalignment with stated purpose.

Key risks include hardcoded credentials, unauthorized network activity, improper tool usage, file I/O vulnerabilities, obfuscated or suspicious code patterns, external data fetches, and credential handling flaws. A thorough audit assesses whether the Skill's actual behavior matches its documented purpose and trust expectations.

Yes. Security auditing covers Skill directories containing Python, Shell, and JavaScript scripts alongside SKILL.md documentation, configuration files, and subdirectories. The audit applies uniformly across diverse file types and usage contexts to detect vulnerabilities spanning network, file I/O, and credential patterns.

After auditing detects vulnerabilities and suspicious patterns, implement mitigation recommendations specific to each risk. Remediation addresses credential exposure, network activity restrictions, code obfuscation removal, and alignment adjustments to ensure the Skill's actual behavior matches its documented safety and security posture.

Security auditing verifies that a Skill's internal behavior—network calls, data handling, credential management, and tool access—aligns with its stated purpose and trust expectations. Pattern detection reveals whether the Skill operates safely or exhibits anomalies indicating misuse or data exfiltration risks.