experience-ui-bundle-metadata-generate

Pass

Audited by Gen Agent Trust Hub on Oct 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local utility scripts included in the package (detect-framework.sh, verify-bundle-location.sh, and check-api-version.sh) to perform environment checks and validate the Salesforce project structure. These scripts use standard tools like jq, grep, and find to inspect and modify project configuration files.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided source code to identify external domains for generating CSP Trusted Site metadata files. While this creates a surface for indirect prompt injection, the instructions include structural constraints for the agent.
  • Ingestion points: User source code is analyzed to extract external URLs (e.g., image CDNs, third-party APIs).
  • Boundary markers: The instructions do not define specific delimiters for the code analysis phase, relying on the agent's ability to extract specific patterns.
  • Capability inventory: The skill can write XML metadata files to the force-app/main/default/cspTrustedSites/ directory and update the sfdx-project.json file.
  • Sanitization: Instructions explicitly direct the agent to extract only the origin (scheme and host), which serves as a filter against processing full URL paths or malicious payloads embedded in URLs.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 3, 2026, 01:29 AM