The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection because it processes user-provided requests to generate or modify demo code.
Ingestion points: User instructions for creating or revising demos under the demos/ directory in SKILL.md.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present.
Capability inventory: The skill directs the agent to execute shell commands using pnpm (including typecheck, test, and lint) within the demo's directory as outlined in the Validation section.
Sanitization: No explicit sanitization or validation of the input is specified beyond the requirement to follow the detailed structural rules.
[COMMAND_EXECUTION]: The instructions require the agent to run local development and validation scripts using the pnpm package manager.
Evidence: The Validation section of SKILL.md explicitly instructs the agent to run pnpm -C demos/<name> typecheck, pnpm -C demos/<name> test, and pnpm run lint to ensure the quality of the artifacts.

make-demo