consumer-goods-tpe-datakit-deploy

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads a static resource package (CGCloudAddons) directly from the user's Salesforce organization using the standard REST API (/services/data/vXX.X/sobjects/StaticResource/.../Body). This is an expected operation for the skill's purpose.
  • [REMOTE_CODE_EXECUTION]: The skill executes a Node.js script (deploy-tpm-tpe.js) that is extracted from the downloaded CGCloudAddons static resource. This execution is performed within the context of the user's authenticated Salesforce environment to complete the deployment.
  • [COMMAND_EXECUTION]: Utilizes execFileSync to run Salesforce CLI commands (sf project deploy, sf data query) for metadata management and system utilities (tar, unzip) for archive handling. It also includes steps to assign permission sets to the user following explicit confirmation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external metadata from the Salesforce organization, specifically XML-based data source definitions, to dynamically calculate required field and object permissions.
  • Ingestion points: *.dataSourceObject-meta.xml files from the downloaded resource, and SOQL query results for deployment logs and existing permissions.
  • Boundary markers: The skill relies on specific code logic and user confirmations rather than NLP-based delimiters to process this data.
  • Capability inventory: File system modification for package management, shell command execution, and network operations via Salesforce APIs.
  • Sanitization: Implements soqlQuote to sanitize variables interpolated into SOQL queries, mitigating injection risks in the database layer.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 09:21 AM
Security Audit — agent-trust-hub — consumer-goods-tpe-datakit-deploy