developing-agentforce
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains templates and instructions designed to make agents more resilient to prompt injection. These include directives for agents to ignore user overrides and to protect internal system configuration from unauthorized exposure.
- [DATA_EXFILTRATION]: Guidance is provided on the use of platform properties like
filter_from_agent. This is a standard Salesforce feature for data minimization, ensuring that technical metadata such as record IDs or internal success flags are not leaked to end-users in the conversation interface. - [COMMAND_EXECUTION]: The skill documents the use of the Salesforce CLI (
sf) for standard development and metadata management tasks. All command patterns are appropriate for the intended developer workflow and restricted to the Salesforce environment.
Audit Metadata