The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external file content using the Gemini CLI's inclusion patterns (e.g., @path and @directory), which creates a risk that malicious instructions embedded in those files could influence the agent's actions.
Ingestion points: SKILL.md and modules/gemini-specifics.md document the use of file path globbing to include content in CLI prompts.
Boundary markers: No specific delimiters or system instructions are provided to the agent to help it distinguish between provided data and embedded instructions in the ingested files.
Capability inventory: The skill enables shell command execution via the gemini CLI and local file system writes to the delegations/ directory.
Sanitization: No sanitization or filtering of file content is performed before interpolation into the prompt context.

gemini-delegation