django-verification
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a wide range of system and application commands (e.g.,
manage.py,pytest,ruff,bandit,pip-audit,git diff) to perform environment checks, security audits, and code quality assessments. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local project environment, including
git diffoutput, project source code during linting, and security scan reports. This content is aggregated into a final summary report. - Ingestion points: Reads project files,
git diffoutput, and tool reports (bandit-report.json,schema.json). - Boundary markers: The skill does not explicitly define delimiters to separate untrusted project content from agent instructions in the output template.
- Capability inventory: The skill has capabilities to read files, execute shell commands, and generate network requests via security tools.
- Sanitization: There is no evidence of sanitization or escaping for the content interpolated into the output report.
- [DYNAMIC_EXECUTION]: Utilizes here-doc blocks (
<< EOF) to pipe dynamically constructed Python scripts into the Django management shell for verifying database indexes, configuration settings, and logging functionality. - [COMMAND_EXECUTION]: Includes an example command for superuser creation that uses a hardcoded password ('admin'). While intended for test environments, this practice can lead to insecure deployments if copied verbatim to production scripts.
Audit Metadata