configuring-connections
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill incorporates robust security practices for handling sensitive data. It explicitly forbids the agent from accepting or repeating real credentials in chat, instead directing users to input secrets through the Celigo UI. It also provides guidance on using placeholders and confirms that the platform masks sensitive fields with '******' during retrieval to prevent accidental exposure.
- [INDIRECT_PROMPT_INJECTION]: The skill manages resources by reading metadata such as connection names and descriptions from the Celigo platform through tools like
list_connectionsandget_schema. While an adversary with platform access could attempt to inject instructions into these metadata fields, the skill provides specific naming standards and clear role definitions to mitigate the risk of the agent misinterpreting data as instructions.
Audit Metadata