zod
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses directive and authoritative language ("CRITICAL", "MUST", "Do not proceed without") to override the agent's internal training data and force it to perform a specific action (fetching a URL) before responding to user queries.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow that ingests external data from the official Zod documentation site. While the source is a well-known service for the library, this creates a surface where instructions in the remote content could influence agent behavior.
- Ingestion points: The agent is instructed to fetch
https://zod.dev/llms.txtviaWebFetchat the start of interactions involving Zod. - Boundary markers: The instructions do not provide delimiters or warnings to ignore potentially malicious instructions embedded in the fetched text.
- Capability inventory: The skill does not explicitly define file-write or subprocess tools, but it operates within the agent's global capability context.
- Sanitization: There is no specified sanitization or filtering for the content retrieved from the external URL.
Audit Metadata