The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to run localized Python scripts, specifically scripts/dependency_mapper.py and scripts/file_stats.py, to perform structural analysis. These scripts are internal to the skill's package and are used for calculating line counts and resolving import graphs.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user source code as part of its architectural review function. While this constitutes a data ingestion surface, the risk is mitigated by the skill's limited capabilities, which are focused on read-only static analysis and local script execution without network exfiltration or persistent file-writing operations.
[CREDENTIALS_UNSAFE]: The skill instructions include detection logic for hardcoded configuration values like URLs and IP addresses within the analyzed project. This is a legitimate architectural audit feature and the skill does not attempt to exfiltrate or store this information.

codeprobe-architecture