The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of untrusted data from external sources such as web URLs and uploaded documents.
Ingestion points: The skill defines API endpoints and provides utility scripts (scripts/source_ingestion.py) for importing external content (PDFs, Office documents, web pages) into the application's knowledge base.
Boundary markers: The instructions do not define delimiters or explicit warnings to prevent the agent from potentially obeying instructions embedded within the ingested content.
Capability inventory: The agent is instructed to summarize, search, and apply custom transformations to the ingested untrusted content.
Sanitization: No sanitization or validation logic is presented in the skill's scripts or documentation for the processed data.
[EXTERNAL_DOWNLOADS]: The installation instructions in SKILL.md recommend downloading a docker-compose.yml file from an external GitHub repository (https://raw.githubusercontent.com/lfnovo/open-notebook/main/docker-compose.yml).
[COMMAND_EXECUTION]: The documentation includes shell commands for environment setup, managing encryption keys, and deploying the application services using docker-compose.
[DYNAMIC_EXECUTION]: The test suite (scripts/test_open_notebook_skill.py) utilizes the Python compile() function to programmatically verify the syntax of example Python scripts provided with the skill. This is a standard and benign use case for script validation in a test context.

open-notebook