elysia
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains a block using high-priority markers ("CRITICAL", "MUST", "Do not proceed") to instruct the agent to disregard its training data and fetch external documentation. This pattern is a form of prompt steering aimed at overriding the agent's default knowledge base.
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest external data via a
WebFetchcall tohttps://elysiajs.com/llms.txt. This creates a vulnerability where instructions embedded in the external document could influence the agent's logic or code generation. - Ingestion points: The skill body in
SKILL.mdtriggers a fetch of external text data. - Boundary markers: There are no instructions provided to treat the fetched content as data only or to ignore embedded commands within the documentation.
- Capability inventory: The skill enables the agent to generate server-side TypeScript code, including routing, validation, and authentication logic.
- Sanitization: No sanitization or schema validation is applied to the content retrieved from the external URL before it is processed by the agent.
Audit Metadata