The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Python to execute its sanitization script as defined in the entrypoint. This is standard behavior for a Python-based skill and does not involve arbitrary or unsanitized shell commands.
[DATA_EXFILTRATION]: No network operations or external data transmissions were detected. The script processes input from stdin and writes output to stdout or local workspace files.
[CREDENTIALS_UNSAFE]: No hardcoded credentials, API keys, or sensitive environment variable accesses were found in the codebase.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute code from external sources. All logic is contained within the local sanitize.py script.
[PROMPT_INJECTION]: The skill's instructions are focused on deterministic text processing. There are no patterns suggesting attempts to override agent behavior or safety filters.
[SAFE]: The skill implements robust security checks for file path resolution in _workspace_file. It prevents directory traversal by resolving paths and ensuring they remain within the cwd. It also explicitly blocks symlinks to prevent symbolic link attacks. File writes are handled atomically using tempfile.mkstemp and os.replace to ensure data integrity.

paper-latex-sanitizer