The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process data from external, potentially untrusted sources including GitHub issues, pull requests, and review threads.
Ingestion points: Instructions specifically direct the agent to 'Harvest every mechanism already proposed in the linked issues, PRs and review threads' and 'Harvest stated constraints'.
Boundary markers: The instructions do not specify any delimiters or safety markers for the agent to use when interpolating this external content into its design analysis.
Capability inventory: The skill allows the agent to create and execute code (prototypes) in the local tmp/ directory and read repository files.
Sanitization: There are no instructions for the agent to sanitize or validate the content harvested from external threads before analysis or prototype generation.
[DYNAMIC_EXECUTION]: The skill requires the agent to generate and execute code at runtime to validate design candidates.
Evidence: The 'Workflow' section instructs the agent to 'Prototype the cheapest candidate in a throwaway under the project's tmp/ and run it against at least three of those fixtures.'
Execution method: This involves script generation and subsequent execution within the local filesystem environment, though restricted to the tmp/ directory.

aidesign