The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted user-supplied documents (PDF, DOCX, TXT, MD) and previews the content to the user. This creates an indirect prompt injection surface where malicious text in a document could attempt to influence the agent's behavior during the reading session preview.
Ingestion points: SKILL.md (Step 1: Parse the Document)
Boundary markers: Absent (no delimiters or instructions to ignore embedded commands are specified)
Capability inventory: Bash, Read, Write tools available
Sanitization: Absent (no evidence of text sanitization or filtering before content is displayed as a preview)
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute local Python scripts (parse_document.py, orp_engine.py). While these are internal scripts, the execution of shell commands based on user-provided file paths requires careful handling to prevent command injection.
[EXTERNAL_DOWNLOADS]: The skill depends on docx and pdfminer. These are well-known, legitimate Python packages used for document parsing and do not represent a significant security risk when sourced from standard registries.

rsvp-reading