The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides procedures and code examples for uploading local files (e.g., CSV, Excel, images) to the protocols.io platform using multipart form-data. While this is a core functionality for managing scientific records, it establishes a pattern for transferring local data to an external service.
Evidence: Found in references/file_manager.md which demonstrates curl -F "file=@/path/to/local/data.xlsx" and Python code in SKILL.md using requests to upload files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to retrieve and process external protocols, experimental steps, and user comments from protocols.io. Since this content is community-contributed, it represents a surface for indirect prompt injection where malicious instructions could be embedded in scientific data.
Ingestion points: Protocol details via GET /protocols/{id}, comments via GET /protocols/{id}/comments, and publications via GET /publications (as documented in references/protocols_api.md, references/discussions.md, and references/additional_features.md).
Boundary markers: The instructions do not specify delimiters or "ignore instructions" warnings to isolate the external protocol content from the agent's core logic.
Capability inventory: The skill documentation includes network operations (REST API calls), local file reading (for uploads), and shell command usage (curl examples).
Sanitization: There are no instructions for the agent to sanitize or validate the ingested protocol text or comments for potentially malicious commands or instructions.

protocolsio-integration