The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data in the form of documents, which constitutes a surface for indirect prompt injection.
Ingestion points: External files are processed via tools such as inspect_document, extract_document_content, and export_document_markdown in SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or clear warnings to the agent to disregard instructions found within the extracted document content.
Capability inventory: The skill has the ability to read local files and write results to local output directories.
Sanitization: The skill description does not outline specific sanitization or filtering procedures for the content extracted from documents.
[COMMAND_EXECUTION]: The skill utilizes the run_sdk_snippet tool to execute dynamically generated Python code for interacting with the document-converter SDK.
The workflow involves constructing Python scripts that import the sdk.tool module to perform document operations.
While the skill provides safe templates for this execution, it represents a mechanism for dynamic code execution to manage file system tasks and document transformations.

document-converter