The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a script scripts/databricks-libs.py that fetches runtime library information from Databricks' official documentation at docs.databricks.com.
[COMMAND_EXECUTION]: The skill uses tmux commands (send-keys and capture-pane) in references/cloud-auth.md to execute cloud commands within a user-authenticated shell pane. This is designed as a safety measure to avoid handling interactive credentials directly within the agent's environment.
[DYNAMIC_EXECUTION]: The skill provides instructions in references/databricks.md for installing the jupyter-databricks-kernel (a vendor-owned resource from the author i9wa4) and executing Jupyter notebooks locally.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface in scripts/databricks-libs.py.
Ingestion points: The script fetches HTML content from the Databricks documentation site (docs.databricks.com).
Boundary markers: None identified in the scraping or output generation process.
Capability inventory: The script generates dependency configuration files (pyproject.toml, requirements.txt) which are then used by the agent to configure local environments.
Sanitization: The script uses regular expressions to extract library names and versions from the HTML source.

data-platform