seed
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and prioritize instructions from files residing within the repository being worked on.
- Ingestion points: In SKILL.md and throughout PLAYBOOK.md, the agent is instructed to read and treat repo-local files like
AGENTS.mdandSEED.mdas authoritative sources of truth for its behavior. - Boundary markers: The playbook does not define clear boundaries or 'ignore embedded instructions' warnings for content loaded from the repository, potentially allowing repository content to override the skill's core logic.
- Capability inventory: The agent is granted significant autonomy in PLAYBOOK.md §1, including the ability to modify project files, initialize git, and add build tools or dependencies.
- Sanitization: There is no requirement for the agent to sanitize or validate the instructions provided in the repository-based files before executing them.
Audit Metadata