The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted project files, such as source code and configuration manifests (e.g., package.json, requirements.txt). This creates an attack surface where malicious content within these external files could attempt to influence the agent's behavior.
Ingestion points: The verifier instructions in references/verifier-ts.md and references/verifier-py.md direct the agent to read various project files including manifests, environment examples, and source code.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the prompts.
Capability inventory: The agent possesses file system read capabilities and the ability to execute shell commands (e.g., npx).
Sanitization: No sanitization or validation of the ingested file content is specified.
[COMMAND_EXECUTION]: The TypeScript verification instructions require the agent to execute a shell command to perform type checking.
Evidence: The file references/verifier-ts.md instructs the agent to "Run npx tsc --noEmit to check for type errors."
[DYNAMIC_EXECUTION]: The skill uses the TypeScript compiler (tsc) via a shell command to analyze the code at runtime.
Evidence: The verification process in references/verifier-ts.md involves running npx tsc against the target application's source code.

sdk-verify