The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/generate_key.py contains a generate_script method that assembles a Bash script using string templates and content passed via command-line arguments. This allows for the creation and execution of arbitrary code based on input parameters.
[PRIVILEGE_ESCALATION]: The generate_key.py script uses file_path.chmod(0o755) to automatically grant execution permissions to the shell scripts it generates, which is a common pattern for privilege escalation or executing payload files.
[INDIRECT_PROMPT_INJECTION]: The skill is intended for auditing codebases and security implementations, which involves processing untrusted external data. If this data is interpolated into the content parameters of the generator script, it could lead to the generation of malicious executables.
Ingestion points: The instructions in SKILL.md specify identifying security scan scopes and targets across codebases.
Boundary markers: No delimiters or warnings are provided to the agent to treat processed data as untrusted.
Capability inventory: The skill possesses file writing capabilities and the ability to modify file permissions.
Sanitization: No input validation or escaping is present in the scripts/generate_key.py script before generating executable content.
[METADATA_POISONING]: There is a discrepancy between the stated purpose of the script scripts/generate_key.py (described as generating encryption keys in its docstring and scripts/README.md) and its actual implementation as a generic markdown, JSON, and shell script generator.

encrypting-and-decrypting-data