The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the databricks CLI and helper scripts (scripts/build_dashboard.py, scripts/ingest.sh) to perform operations such as creating tables, dashboards, and apps. These executions are scoped to the Databricks environment and are essential for the skill's functionality.
[EXTERNAL_DOWNLOADS]: Documentation within the skill references the author's official GitHub repository (github.com/Nimbleway/cookbook) for setup and integration purposes. These are trusted vendor resources.
[PROMPT_INJECTION]: The skill ingests data from external websites via Nimble agents. To prevent indirect prompt injection, the skill instructions specify the use of robust SQL sanitization patterns, such as try_cast and regexp_replace, ensuring that scraped content is safely processed into structured formats.
Ingestion points: External data is ingested via the nimble_agent_run SQL function call in Phase 3 of SKILL.md.
Boundary markers: Content is parsed and stored into predefined Delta table schemas as described in references/nimble-agents.md.
Capability inventory: The skill performs SQL operations and makes API calls to manage Databricks resources across various scripts and reference files.
Sanitization: The skill employs SQL-based cleaning and type-safe casting to validate input data, notably in the ingest SQL template in references/nimble-agents.md.

nimble-databricks-data-products