The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines workflows that ingest untrusted data from pull requests and local git repositories to provide context for the AI agent.
Ingestion points: Data from gh pr view (titles, bodies, files) and git diff is interpolated into prompts for the jules new command.
Boundary markers: None; external content is included directly in the task description without delimiters.
Capability inventory: The jules CLI has the ability to apply code changes locally, clone repositories, and interact with the remote Jules service.
Sanitization: The ingested data is not validated or escaped before being processed by the AI.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill instructs the user to install a specific package from the npm registry.
Evidence: npm install -g @google/jules
Context: The package is an official tool for the Google Jules service and is essential for the functionality described in the skill.

jules