The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to analyze untrusted source code, which creates a potential vector for indirect prompt injection where malicious instructions embedded in the code could influence the agent's behavior.
Ingestion points: The scripts scripts/pr_analyzer.py, scripts/code_quality_checker.py, and scripts/review_report_generator.py are designed to take a project path as input for analysis.
Boundary markers: There are no explicit delimiters or system instructions provided within the scripts to differentiate between code data and potential instructions.
Capability inventory: The scripts include a file-writing capability via the --output argument, allowing the agent to save analysis results to the local file system.
Sanitization: The current implementation lacks sanitization logic to strip or escape potential injection patterns from the analyzed source code before processing or reporting.

code-reviewer