The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the Stitch Effective Prompting Guide at https://stitch.withgoogle.com/docs/learn/prompting/. This refers to an official documentation page from a well-known service (Google).
[INDIRECT_PROMPT_INJECTION]: The skill ingest untrusted user input (vague UI ideas) and local project files (DESIGN.md) to generate structured prompts for a downstream system (Stitch).
Ingestion points: User input prompts and the content of DESIGN.md (read from the local environment).
Boundary markers: The skill lacks explicit boundary markers or instructions to the LLM to ignore potentially malicious embedded commands within the design content or user input.
Capability inventory: The skill utilizes Read and Write tools to access project files and potentially output the results to local files like next-prompt.md.
Sanitization: No specific sanitization or filtering logic is present to prevent the propagation of instructions hidden within the UI ideas into the final generated prompt.

enhance-prompt