The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by reading Excel, Word, PowerPoint, and PDF documents and converting them into text formats for the AI assistant.
Ingestion points: The office_read, office_list_pdf_fields, and office_analyze_pdf_layout tools are designed to ingest data from user-provided file paths.
Boundary markers: There are no instructions or delimiters provided to warn the AI assistant to ignore instructions that might be embedded within the documents it processes.
Capability inventory: The skill provides powerful file-system capabilities, including creating new documents (office_create_xlsx), modifying existing ones (office_replace_text), and exporting to PDF (office_export_pdf), which could be abused if the agent follows malicious instructions from a document.
Sanitization: No sanitization or content validation is mentioned for the data extracted from documents before it is returned to the AI's context.
[EXTERNAL_DOWNLOADS]: The skill's installation process involves downloading and executing code from external repositories.
Evidence: The instructions guide the user to perform cargo install office-oxide-mcp or git clone https://github.com/Aimino-Tech/office-oxide-mcp.git to obtain the core MCP server binary.
The binary is hosted on GitHub and the Cargo registry, which are standard platforms for software distribution, and it executes with local user permissions to provide the document processing functionality.

opendocswork-mcp-office-processing