cloudbase-platform
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill leverages official management tools such as
manageEnv,manageGateway,manageFunctions, andmanagePermissionsto configure platform resources. These operations are within the expected scope of a platform management skill. - [EXTERNAL_DOWNLOADS]: It references official vendor packages, including
@cloudbase/js-sdk,@cloudbase/node-sdk, and@cloudbase/wx-cloud-client-sdk. It also utilizessearchKnowledgeBaseto fetch information from official documentation, with no evidence of untrusted remote script execution. - [INDIRECT_PROMPT_INJECTION]: While the skill ingests external data (documentation and environment info), it includes a dedicated
Sensitive Runtime Data Protectionprotocol. This protocol provides explicit instructions to prevent the agent from echoing sensitive platform context (likex-cloudbase-context) or credentials back to the user, effectively mitigating output-based injection and leakage risks. - [SAFE]: The inclusion of mandatory pre-check protocols like the
Deployment Gateand theChange Safety Protocoldemonstrates a high level of security awareness. These protocols enforce user confirmation and verification steps for all non-trivial operations, reducing the risk of accidental misconfiguration or unauthorized deployments.
Audit Metadata