The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads and analyzes existing source code files to identify blockers or perform automated code translations (Section 5).
Ingestion points: The agent reads external source code files using search tools like grep or rg (SKILL.md, Section 5).
Boundary markers: The instructions lack specific delimiters or warnings to ignore instructions that might be embedded within the external code files.
Capability inventory: The skill can write new files (e.g., <name>.classic.py), modify job configurations (performance modes, environments), and trigger autonomous deployments.
Sanitization: No sanitization or validation steps are defined for content read from external files before it is interpolated into deployment specifications.
[COMMAND_EXECUTION]: The skill utilizes the databricks CLI and system tools like grep or rg. The instructions encourage the agent to deploy changes autonomously ("Do not stop to ask whether to migrate. Deploy, then report what you did"), which bypasses user confirmation for the generated configurations.

databricks-serverless