The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script check.py uses importlib.import_module() to verify the presence of Python dependencies identified during its scan.
Evidence: The function is_importable(module) in check.py passes module names extracted from scanned files directly to importlib.import_module().
Risk: In Python, importing a module triggers the execution of any code defined at the module's top level. If the checker scans a script that references a malicious local module or a hijacked package name, that code will execute in the context of the checker.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface common to static analysis tools that process external, untrusted source code.
Ingestion points: In check.py, the skill reads the content of companion scripts (.py, .sh) and SKILL.md files from directories defined in SUPERPOWERS_DIR.
Boundary markers: There are no markers or sanitization logic to distinguish between legitimate code and malicious sequences designed to exploit the linter's logic.
Capability inventory: The checker has the capability to read files and trigger code execution via the Python import system.
Sanitization: The module names are extracted using a regular expression (IMPORT_RE) and used without further validation, allowing an attacker to control the input to a dynamic execution sink.

skill-portability-checker