The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process various untrusted external data sources to clarify project requirements.
Ingestion points: Found in SKILL.md (confidential working files) and references/existing-artifact.md (mocks, designs, spreadsheets, sample data files, reference links, and recordings).
Boundary markers: The agent is instructed to "Say back what you see" and "read the key decisions back as one short list for a yes" before committing changes, creating a human-in-the-loop verification step.
Capability inventory: The skill possesses the ability to write to the masterplan file and generate executable code for behavior and structure prototypes.
Sanitization: The instructions explicitly mandate using fake or disposable data and prohibit extracting real secrets or passwords from artifacts like screenshots.
[DYNAMIC_EXECUTION]: The skill directs the agent to generate "coded throwaway" prototypes to test application logic and layout.
Evidence: references/prototype-behaviour.md and references/prototype-structure.md instruct the agent to build standalone files (e.g., HTML/JS) that can be executed in a browser. While intended for local testing, this involves generating executable logic derived from untrusted requirement descriptions and "invented data that looks like theirs."

clarify