The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The skill is designed to orchestrate workflows by executing shell commands and local CLI tools such as gemini, claude, and npx. This is the intended primary purpose of the skill.
[EXTERNAL_DOWNLOADS] (LOW): The scripts utilize npx to execute eslint and tsc. By default, npx may download these packages from the npm registry if they are not found in the local environment.
[PROMPT_INJECTION] (LOW): (Category 8
Indirect Prompt Injection) The skill exhibits a significant attack surface for indirect prompt injection by interpolating untrusted data into LLM prompts.
Ingestion points: Command-line arguments ($INPUT, $TOPIC) and local file contents (read via cat "$FILE") are used as variables.
Boundary markers: None. The skill does not use delimiters (like XML tags or triple backticks) or instructions to the LLM to ignore embedded commands within the variables.
Capability inventory: The pipeline has the capability to read files (cat), execute shell logic, and run node packages (npx).
Sanitization: There is no evidence of sanitization, escaping, or validation of the input data before it is passed to the LLM engines.

pipeline