The Agent Skills Directory

[SAFE]: The skill purpose is session context management using a tiered file structure in a local memory/ directory. No malicious code, obfuscation, or unauthorized access patterns were identified across the 6 analyzed files.
[PROMPT_INJECTION]: The skill handles potential indirect prompt injection by implementing a provenance model for project decisions. It specifically instructs the agent to distinguish between user-approved decisions and inferred metadata, treating the latter as advisory to prevent untrusted data from overriding project configuration.
[DATA_EXFILTRATION]: No unauthorized network operations or exfiltration patterns were found. The skill includes built-in privacy compliance procedures, such as data retention policies and a purge mechanism for third-party personal data, demonstrating security-conscious handling of project information.

memory-management