sca-pip-audit
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Indirect Prompt Injection] (MEDIUM): The skill processes
requirements.txtfiles, which are untrusted external inputs. A maliciously crafted requirements file could attempt to influence the agent's behavior or cause it to suggest installing compromised versions of packages during the remediation step. - Ingestion points:
requirements.txtviapip-audit -r. - Boundary markers: None present in the instructions to separate untrusted data from instructions.
- Capability inventory: Execution of
pip-auditandpip installsubprocesses. - Sanitization: No explicit sanitization or validation of the requirements file content is described.
- [Command Execution] (LOW): The skill explicitly uses shell commands (
pip-audit,pip install). While these are intended for security maintenance, they allow modification of the local environment. - [External Downloads] (LOW): The skill requires the installation of
pip-auditfrom PyPI. Per the [TRUST-SCOPE-RULE], this is a trusted source, but the dependency itself is necessary for the skill's operation.
Audit Metadata