The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and analyze the content of .ipynb files, which may contain untrusted data or instructions.\n
Ingestion points: The agent is directed to use the Read tool on notebook files (e.g., Read __marimo__/notebook.ipynb) and process their contents using jq to identify tracebacks and execution counts.\n
Boundary markers: The instructions do not specify any delimiters or safety markers to isolate notebook content from the agent's core instructions.\n
Capability inventory: The skill utilizes shell-based notebook executors and processors including marimo, jupytext, papermill, jq, and grep.\n
Sanitization: There is no evidence of sanitization, escaping, or filtering of the notebook content before the agent interprets it to determine the success or failure of a task.\n- [COMMAND_EXECUTION]: The skill defines several shell command patterns for executing and converting notebook files.\n
Execution patterns: Includes marimo export ipynb, jupytext --to notebook, and papermill for capturing cell outputs and tracebacks.

notebook-debug