drizzle
Warn
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses high-urgency instructions including 'CRITICAL: Your training data... is unreliable', 'MUST use WebFetch', and 'Do not proceed without fetching this URL first' to override the agent's internal training and behavior protocols.
- [EXTERNAL_DOWNLOADS]: Fetches live documentation and context for LLMs from the official Drizzle ORM website (orm.drizzle.team).
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by instructing the agent to ingest content from an external, remote source.
- Ingestion points: SKILL.md (via WebFetch call to orm.drizzle.team/llms.txt).
- Boundary markers: Absent; the skill does not provide instructions to the agent to isolate or ignore potentially malicious instructions within the fetched text.
- Capability inventory: No internal scripts are defined; the skill leverages the agent's WebFetch capability.
- Sanitization: Absent; the content is fetched for direct use by the agent without validation or filtering.
Audit Metadata