platform-sharing-owd-configure

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Salesforce CLI (sf) to execute queries and metadata operations. It builds shell commands by directly inserting user-provided values like '' and '' into command strings, such as in the sf data query and sf project retrieve operations. This pattern creates a risk of command injection if the user provides crafted strings. \n- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and updates .object-meta.xml metadata files from a Salesforce organization. These files represent an ingestion point for untrusted data that could contain malicious instructions aimed at the agent. \n
  • Ingestion points: User-supplied object names and .object-meta.xml files retrieved via the CLI. \n
  • Boundary markers: No specific delimiters or 'ignore' instructions are present when interpolating these values into commands or processing file content. \n
  • Capability inventory: The skill can query data, retrieve files, and deploy changes to the target Salesforce environment. \n
  • Sanitization: The instructions do not define sanitization or escaping logic for object names used in shell commands. \n- [SAFE]: All identified tools (sf) and resource patterns are consistent with the Salesforce vendor ('forcedotcom') and the skill's stated administrative purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 10:35 PM
Security Audit — agent-trust-hub — platform-sharing-owd-configure