The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a code auditing tool (scripts/code_quality_analyzer.py) that ingests untrusted source code files from the local filesystem to perform security and complexity scans. This ingestion represents a potential attack surface where malicious instructions embedded in the analyzed code could attempt to influence the agent's behavior during the reporting phase.
Ingestion points: The script scripts/code_quality_analyzer.py recursively reads all files within a user-specified directory using Path.rglob("*") and open().read().
Capability inventory: The skill possesses file reading and writing capabilities (e.g., writing reports via the --output flag), but does not contain network operations, subprocess execution, or arbitrary command execution logic.
Boundary markers: The analyzer treats file content as static data for regex-based pattern matching; the isolation between the analysis script and the data is considered sufficient for its intended purpose.
Sanitization: The script performs pattern matching to identify issues; however, it does not explicitly sanitize the source code snippets it might include in a report before they are presented to the agent context.

senior-fullstack