The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze "surrounding code" in the user's local environment to determine the appropriate prototyping strategy. This ingestion of potentially untrusted data from the workspace provides a surface for indirect prompt injection attacks.
Ingestion points: SKILL.md instructs the agent to identify context from "surrounding code" (e.g., backend modules vs. UI components) to decide between logic or UI prototype branches.
Boundary markers: The instructions do not specify the use of delimiters or explicit "ignore" warnings when processing the workspace code context.
Capability inventory: The skill utilizes shell execution capabilities (pnpm, python, bun) and has permission to write files throughout the project directory.
Sanitization: There are no documented steps for sanitizing or validating the code context ingested from the local files before it is processed by the agent.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to run the generated prototypes via task runners. SKILL.md explicitly lists pnpm, python, and bun as tools for launching the code within the project environment.

prototype