The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs an automated installation of the pdf-parse library from the npm registry using npm install during the environment setup phase.
[COMMAND_EXECUTION]: Executes a local Node.js script (extract_pdf_text.js) via the command line to process user-provided PDF files and extract their text content.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted input to generate executable code.
Ingestion points: Reads content from user-provided .pdf, .txt, and .md files in Step 1 and Step 2 of SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used when passing the extracted text to the code generation phase.
Capability inventory: The skill uses the Write tool to create .html and .js files containing executable scripts (Step 4 and Step 3.5).
Sanitization: There is no evidence of sanitization or escaping of the input text before it is interpolated into the HTML/JS generation logic, which could allow an attacker to inject malicious scripts into the resulting prototype.

html-prototype