The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data retrieved through web search tools and the Context7 MCP documentation tools.
Ingestion points: External content enters the agent's context via the #websearch tool and the get-library-docs MCP tool (SKILL.md).
Boundary markers: The instructions include a "Verification Protocol" to define agent boundaries, scope delegated tasks, and verify return artifacts (SKILL.md).
Capability inventory: The skill possesses the capability to generate and execute Python code, manage the file system (e.g., creating requirements.txt), and perform static analysis on local files (SKILL.md, scripts/quality-gate.py).
Sanitization: The skill provides explicit instructions for input sanitization and recommends the use of libraries such as bleach for HTML content (SKILL.md).
[EXTERNAL_DOWNLOADS]: The skill's documentation and examples suggest installing various well-known and established Python packages from standard registries.
Evidence: References to pip install graphviz, httpx, pydantic, and security tools like pip-audit are found throughout the documentation (examples/python-diagram-generation-patterns.md, references/python-libraries-guide.md).
[COMMAND_EXECUTION]: The workflow involves the execution of shell commands for project environment management and code quality assurance.
Evidence: Commands for pip freeze, venv creation, and running the provided quality-gate.py script are documented (SKILL.md, scripts/quality-gate.py).

codexer