The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides a command to download and execute a shell script directly from the vendor's official domain for installation and updates.
Evidence: curl -LsSf https://getbruin.com/install/cli | sh in workflow.md.
[EXTERNAL_DOWNLOADS]: The skill fetches an installation script and references documentation from the vendor's web server.
Evidence: Downloads from https://getbruin.com/install/cli and documentation links to getbruin.com.
[COMMAND_EXECUTION]: The skill uses local shell commands and CLI tools to initialize projects, manage git repositories, and configure MCP servers.
Evidence: Execution of bruin, git, and claude mcp add bruin commands throughout the workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external analytics platforms (GA4 and Search Console), which creates a potential surface for instructions embedded in external data to influence the agent.
Ingestion points: GA4 BigQuery exports, GA4 API, and Search Console bulk exports as noted in SKILL.md.
Boundary markers: While it instructs the agent to state data sources and windows, it lacks specific delimiters or escaping for the raw data ingested.
Capability inventory: The skill has the ability to run shell commands, write configuration files, and perform network operations via the Bruin CLI.
Sanitization: Includes rules to strip sensitive query strings from URLs and warns that IP/user-agent data should not be considered fully anonymous.

web-analytics