The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The resource-lister.py script is configured to extract and report the value of the DATABASE_URL environment variable. This is a high-risk finding as this variable is the standard location for database connection strings which often include sensitive credentials in plain text.
[DYNAMIC_EXECUTION]: The script utilizes the __import__() function within check_package_installed to dynamically load Python modules. Since the module names can be supplied by a user or the agent via the --check command-line argument, this presents a risk where importing a malicious or unexpected module could trigger arbitrary code execution at the module's top-level scope.
[COMMAND_EXECUTION]: The script uses subprocess.run() to execute numerous system commands, including package managers (pip3), database clients (psql, redis-cli), and various engineering tools. This provides a broad surface for interacting with the underlying operating system shell.
[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to provide environmental context to the agent to influence its future actions (e.g., determining which libraries to use in code generation). Because the script ingests unsanitized data from the environment—such as software version strings, environment variable values, and file paths—it creates a surface where a malicious environment could inject instructions to manipulate the agent's behavior.

get-available-resources