The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides an installation command for the Bruin CLI using curl -LsSf https://getbruin.com/install/cli | sh in workflow.md. This is a standard installation pattern from the official vendor website.
[COMMAND_EXECUTION]: The skill utilizes several local CLI tools including bruin (init, version, upgrade, run, validate) and git to manage data pipelines and project setup. It also includes instructions for adding an MCP server via claude mcp add.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from sources like PostHog, Firebase, and various template configuration files.
Ingestion points: The agent is instructed to read pipeline.yml variables, asset filters, and template READMEs as described in workflow.md.
Boundary markers: Guidelines in analysis.md instruct the agent to verify data freshness and source documentation before processing or writing SQL.
Capability inventory: Includes the ability to execute bruin run, bruin validate, and file system operations via git.
Sanitization: Instructions in SKILL.md explicitly forbid pulling raw user-level event streams into context and mandate aggregation in SQL first to minimize data exposure.

product-analytics