gardeners
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill spawns and coordinates sub-agents that process project files and external data, creating a potential surface for indirect prompt injection where instructions embedded in the codebase could influence agent behavior.
- Ingestion points: Project files and issue descriptions are ingested by sub-agents during the gardening process.
- Boundary markers: The provided prompt template for sub-agents does not include explicit delimiters or instructions to disregard potential commands found within the scanned files.
- Capability inventory: Sub-agents have the ability to spawn additional agents, manage a shared task list, send messages to other agents, and create pull requests (via the referenced garden skill).
- Sanitization: The instructions do not specify any sanitization or validation steps for the content scanned and processed by the sub-agents.
Audit Metadata