The Agent Skills Directory

Standard Package Installation: The skill suggests installing well-known Python packages (jupyterlab, nbformat, nbclient, ipykernel) to support notebook operations. These are standard components of the Jupyter ecosystem.
Notebook Execution Command: The instructions include a command to execute notebooks using jupyter nbconvert. This is a standard method for validating that an analysis artifact is reproducible and free of runtime errors.
Data Analysis Surface (Indirect Prompt Injection): As an analysis-focused skill, it is designed to ingest and process data from external sources such as tables, logs, and query results. While processing external data is a common vector for indirect prompt injection, the skill promotes safety by emphasizing result validation, reasonableness checks, and clear documentation of data sources.
Ingestion points: Data enters the agent context through tools like ~~structured_data and ~~operations_logs, as well as manually referenced source paths and query artifacts.
Boundary markers: The skill encourages logical separation of code and data but does not explicitly require specific boundary delimiters for ingested data strings.
Capability inventory: The primary capability is shell command execution for notebook conversion and execution via jupyter nbconvert.
Sanitization: The skill emphasizes manual review, matching takeaways to executed outputs, and adding local reasonableness checks for surprising results.

jupyter-notebooks