The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements automated code assessment logic that reads external files, creating an indirect prompt injection surface. \n
Ingestion points: The LegacyCodeAnalyzer class in references/system-assessment.md recursively identifies and reads the content of all .py files within a provided codebase path. \n
Boundary markers: There are no explicit instructions or delimiters used when processing file content to warn the agent against following embedded instructions. \n
Capability inventory: The skill allows the agent to generate complex migration plans, write implementation code, and execute local git commands (as seen in system-assessment.md). \n
Sanitization: File content is parsed via the ast module for static analysis, but no sanitization of the raw text is performed prior to being processed by the agent.
[COMMAND_EXECUTION]: The assessment script in references/system-assessment.md utilizes subprocess.run to execute the git log command. This is used legitimately to identify 'hotspots' (frequently changed files) in the legacy system's version history, though it represents a direct shell command execution capability.

legacy-modernizer