The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data, which presents an indirect prompt injection attack surface as defined in Category 8.
Ingestion points: Untrusted data enters the context via the materialText, materialFile, and materialUrl parameters in references/form-schema.json.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands were found in the skill instructions.
Capability inventory: The skill performs network requests via the urllib.request module in scripts/run.py to send data to the backend API.
Sanitization: No sanitization or escaping of the user-provided document content is performed in the local runner script before transmission.
[COMMAND_EXECUTION]: The skill includes an executable Python script scripts/run.py intended for users to run locally to facilitate interaction with the AI Skills API.
[EXTERNAL_DOWNLOADS]: The runner script initiates network connections to the vendor's domain ai-skills.ai. Per the author context guidelines, this domain is recognized as the official endpoint for the skill's operations.

ai-docx