The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's installation guide instructs the agent to download a docker-compose.yml file from the lfnovo/open-notebook GitHub repository. While this is standard for self-hosted software, it involves fetching configuration from an external source that is not part of a verified trusted organization list.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted content (PDFs, web pages, audio/video), which presents a surface for indirect prompt injection attacks.
Ingestion points: SKILL.md and references/api_reference.md describe endpoints for ingesting URLs, files, and raw text into notebooks for AI analysis.
Boundary markers: The documentation does not specify the use of prompt delimiters or directives to ignore instructions embedded within ingested documents when the AI processes the content.
Capability inventory: The skill utilizes the requests library in Python to interact with the local Open Notebook API, supporting actions like file creation, chat execution, and content transformations using the ingested data as context.
Sanitization: There is no evidence of content sanitization or validation of the ingested materials within the skill's instructions.

open-notebook