The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external file content which could contain malicious instructions designed to influence the agent.
Ingestion points: upload_file reads user-selected file bytes (SKILL.md).
Boundary markers: Instructions require explicit user selection and confirmation of file metadata (filename, MIME type, purpose) before processing (SKILL.md).
Capability inventory: Includes file upload, listing, retrieval, sharing link creation, and deletion via MCP tools (SKILL.md, agents/openai.yaml).
Sanitization: Content is encoded into Base64 for transit rather than being interpreted as text within the conversation context (SKILL.md).
[SAFE]: The skill uses the vendor's official MCP infrastructure at mcp.inttegro.com for authenticated file operations.

inttegro-files