consumer-goods-promotion-bo-api-deploy
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFE
Full Analysis
- [DYNAMIC_EXECUTION]: The skill generates and deploys custom Apex classes and anonymous Apex scripts based on user input from an interactive interview. This behavior is the primary purpose of the skill and is implemented with security controls, including a namespace-agnostic template and a mandatory user review step where the generated source code is presented for confirmation before deployment to the Salesforce organization.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data to define deployment parameters and logic. To mitigate injection risks, it dynamically derives the expected data contract directly from the target organization's metadata (using the BO_API_Step_Input_Structure__c schema) and performs comprehensive pre-flight validation of all paths and reference data before any ingestion or deployment occurs.
- [COMMAND_EXECUTION]: Standard Salesforce CLI tools (sf) are used to perform queries and metadata operations. The skill adheres to security best practices by utilizing the CLI's internal authentication mechanisms and explicitly instructing the agent to never extract or store session tokens or access tokens in logs or shell output.
- [PRIVILEGE_ESCALATION]: The skill includes checks for required permissions, such as the TPM Master Data Admin persona and relevant Permission Set Licenses. If missing, it provides instructions for assigning them rather than attempting unauthorized escalation, ensuring the user maintains control over access rights.
Audit Metadata