The Agent Skills Directory

[SAFE]: The skill performs static analysis using a Python script to identify cryptographic vulnerabilities. No malicious patterns, exfiltration attempts, or unauthorized access were detected.
[INDIRECT_PROMPT_INJECTION]: The tool ingests untrusted data by reading source code files from the local filesystem. Malicious instructions could be embedded within code snippets that the scanner identifies, potentially influencing the agent when it reviews the tool's findings.
Ingestion points: The scan_file function in scripts/scan_weak_crypto.py reads file content using Path.read_text.
Boundary markers: The skill output provides the file name and line number, but does not explicitly wrap snippets in safety delimiters.
Capability inventory: The skill is restricted to reading the filesystem, performing glob/grep operations, and executing the provided Python script. It lacks network access and write permissions, limiting the impact of any successful injection.
Sanitization: No sanitization is performed on the extracted code snippets before they are displayed.

detecting-weak-cryptography