The Agent Skills Directory

[SAFE]: No malicious patterns, data exfiltration, or unauthorized privilege escalation were detected. The skill operates locally on the provided codebase using standard search utilities.
[COMMAND_EXECUTION]: The skill provides utility scripts (scan-patterns.sh, trace-data-flows.sh) that execute ripgrep and python3 to automate the discovery of security-relevant patterns. These operations are restricted to the local filesystem and the target project directory.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from codebases (e.g., comments, README files). A malicious actor could attempt to influence the agent's analysis by embedding instructions in the source code. This is a low-severity risk inherent to the skill's purpose, which is addressed by the methodology's focus on evidence-backed data flow traces.

threat-model