The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided application requirements, conversational "vibes," and technical error logs to provide architectural advice and code generation guidance. This behavior establishes an ingestion surface for untrusted external data.
Ingestion points: Technical requirements in natural language and error stack traces provided by the user (found in SKILL.md and references/prompt-patterns.md).
Boundary markers: The instructions recommend asking for confirmation and explaining intent, but do not specify technical delimiters to isolate user-provided content from the agent's core instructions.
Capability inventory: The skill utilizes the agent's core language modeling, web search (to find tool documentation), and code interpretation/generation capabilities.
Sanitization: While the skill provides extensive guidelines for the user to implement sanitization in the generated software, it does not include specific validation or filtering mechanisms for the agent's processing of the user input itself.

vibe-coding