service-agentforce-human-escalation-configure
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses Bash and Node.js scripts to execute
sfCLI commands for querying and deploying Salesforce metadata and data. This is the core functionality of the skill. The execution is protected by regex validation for all developer-supplied identifiers (agents, queues, topics) to prevent command or SOQL injection. - [INDIRECT_PROMPT_INJECTION]: The skill analyzes Agent Script files (AiAuthoringBundle) to verify the presence of escalation actions. This analysis is performed using static regex scanning and comment stripping in Node.js, rather than runtime execution of the scripts, which mitigates risks associated with processing untrusted agent content.
- [SAFE]: The skill implements a mandatory production write-guard that blocks execution on non-sandbox/trial environments. It also uses proper XML entity escaping and shell-safe variable interpolation when generating deployment assets from templates.
Audit Metadata