The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted OpenAPI 3.1+ specifications from files, URLs, or raw text and uses them as the source of truth for generating Python code.
Ingestion points: User-provided OpenAPI documents provided in SKILL.md.
Boundary markers: The instructions do not specify sanitization or escaping of metadata (e.g., summary, description, operationId) when they are inserted into generated code as names or docstrings.
Capability inventory: The skill has the capability to write and modify several project files, including views.py, urls.py, serializers.py, and package manifests like pyproject.toml.
Sanitization: No specific sanitization logic is mentioned for preventing the embedding of malicious instructions within the generated codebase.
[COMMAND_EXECUTION]: The skill provides the agent with instructions to execute environment setup and management commands.
Evidence: Instructions to use package managers like uv, poetry, or pip to install dependencies and python manage.py runserver to start the application in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill fetches framework documentation and suggests installing external packages.
Evidence: References to external documentation at django-modern-rest.readthedocs.io and instructions to install the django-modern-rest library and its extras in SKILL.md.

dmr-openapi-skeleton