The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted project data including source code, dependency manifests (requirements.txt, package.json), and database metadata.\n
Ingestion points: Project files are ingested via tools such as ruff, bandit, pytest, and git diff as described in SKILL.md.\n
Boundary markers: There are no explicit delimiters or instructions to the agent to disregard instructions that might be embedded in the code or data being analyzed.\n
Capability inventory: The skill possesses extensive capabilities including arbitrary shell command execution (python manage.py), file system inspection, and network-enabled dependency scanning.\n
Sanitization: The skill does not implement sanitization or validation of the data read from the project environment before it is processed or displayed in reports.\n- [COMMAND_EXECUTION]: The skill makes extensive use of shell commands to automate the verification pipeline. While these are standard development practices, they represent a significant capability tier that could be exploited if combined with malicious input.\n- [DYNAMIC_EXECUTION]: The skill employs dynamic Python execution through shell redirection (python manage.py shell << EOF) and one-liners (python -c). These patterns are used for environment validation and creating administrative users in a test context.

django-verification