The Agent Skills Directory

[SAFE]: The skill uses established Python and JavaScript libraries for PDF processing. All file operations are restricted to paths provided via command-line arguments, and no network exfiltration patterns were identified.
[SAFE]: The script scripts/fill_fillable_fields.py includes a runtime monkeypatch of the pypdf library. This is a documented workaround for a specific bug in pypdf's handling of selection lists and does not involve untrusted input or malicious redirection.
[PROMPT_INJECTION]: The skill documentation in forms.md uses imperative language and markers like 'CRITICAL: You MUST'. These are instructional for the agent's workflow rather than attempts to bypass safety filters or jailbreak the model.
[PROMPT_INJECTION]: As the skill processes and extracts text from external PDF files, it possesses a surface for indirect prompt injection. Malicious instructions embedded within a PDF document could potentially be interpreted by the agent during processing. This is a known risk for document-parsing tools and is managed by the agent's internal safety protocols.

pdf