tidb-cloud-zero
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions involve querying database tables, which creates an ingestion point for untrusted data that could potentially contain malicious instructions.
- Ingestion points: Database query results returned from SELECT statements described in
SKILL.mdand reference files. - Boundary markers: The skill does not provide specific delimiters or warnings to the agent for handling data returned from queries.
- Capability inventory: No custom scripts or subprocess execution logic are included; the agent utilizes its native SQL capabilities.
- Sanitization: No explicit sanitization or filtering of database content is requested or implemented.
- [SAFE]: All network communications are directed toward authorized vendor infrastructure (e.g.,
zero.tidbapi.com). The skill explicitly warns the user to use TLS and secure credential handling methods.
Audit Metadata