The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/install.py helper facilitates the installation of the pdf-inspector library and its associated Node.js and Rust bindings. These installations use standard package managers (pip, npm, cargo) to fetch official releases of the pdf-inspector tool from Firecrawl.
[COMMAND_EXECUTION]: The scripts/pdf_read.py script supports an optional OCR fallback via the --ocr-cmd flag, which executes a shell command for pages lacking embedded text. The script implements a _shell_quote function that applies platform-specific quoting (e.g., shlex.quote on POSIX) and rejects potentially dangerous characters on Windows (like % and !) to prevent command injection from malicious file names or metadata.
[SAFE]: The skill includes an output sanitization layer (_sanitize) that strips terminal control characters (ANSI escape sequences), C0/C1 controls, and Unicode line separators from the extracted PDF text. This prevents a malicious document from attempting to manipulate the agent's terminal or environment through smuggled escape sequences.

pdf-reader