platform-headless-360-integrate

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements exemplary security practices, including a strict 'read-before-write' policy, mandatory user confirmation for any non-GET requests (even when labeled as reads in recipes), and deterministic API version selection via a dedicated validation script.
  • [COMMAND_EXECUTION]: The skill utilizes local Python scripts (h360_api_version.py, codex_mcp_launcher.py) that are part of the vendor's plugin distribution. These scripts perform benign operations such as validating API version responses and managing local user preferences for the MCP server connection.
  • [INDIRECT_PROMPT_INJECTION]: The skill documents the risk of indirect prompt injection from external 'recipe' data. It provides the agent with explicit instructions to treat all recipe metadata and guidance as untrusted, specifically forbidding the execution of scripts named in that data or following instructions embedded within it.
  • [DATA_EXPOSURE_&_EXFILTRATION]: Network operations are restricted to Salesforce's official 'api.salesforce.com' domains. The skill explicitly instructs against handling or transmitting sensitive credentials like tokens or client secrets within the chat context.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 04:18 PM