The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/type-check.sh is designed to execute the TypeScript compiler (tsc) to validate code quality and type safety. It intelligently searches for the compiler in the system path, local node_modules, or executes it via npx.
[EXTERNAL_DOWNLOADS]: The inclusion of npx tsc in the helper script may result in the download of the official TypeScript package from the public npm registry if the package is not already cached or installed locally. This is a standard behavior for modern JavaScript/TypeScript development environments.
[INDIRECT_PROMPT_INJECTION]: Because the skill is designed to ingest and process TypeScript source code and compiler outputs from user-provided project directories, it possesses an inherent attack surface for indirect prompt injection. Maliciously crafted code comments or string literals within a processed project could attempt to influence the agent's reasoning, though no specific exploits are present in the skill itself.
Ingestion points: The script runs tsc on files within the PROJECT_DIR specified at runtime (found in scripts/type-check.sh).
Boundary markers: Absent; the agent receives the raw compiler error samples encapsulated in a JSON structure.
Capability inventory: The skill can execute shell commands via scripts/type-check.sh, including find, grep, and tsc.
Sanitization: The script performs basic sanitization by escaping double quotes in the ERROR_SAMPLE output to maintain valid JSON structure.
[DATA_EXPOSURE_AND_EXFILTRATION]: The helper script accesses project-specific files such as tsconfig.json and performs directory traversals using find to count TypeScript files. These operations are scoped to the project directory provided to the script and are necessary for the skill's stated purpose of type analysis.

typescript-master