skills/adobe/skills/run-workflow/Gen Agent Trust Hub

run-workflow

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Orchestrates complex Adobe Firefly workflows by calling custom MCP tools such as run_workflow_submit, compose_workflow, and inspect_run to process media and manage workflow state.
  • [EXTERNAL_DOWNLOADS]: Facilitates asset transfer via presigned URLs (Azure SAS, AWS S3) using the upload_asset and download_output tools, which is necessary for cloud-based image and video processing.
  • [SAFE]: Implements secure file system interaction by utilizing the request_folder_access tool to resolve macOS EPERM errors through a native folder picker, following the principle of least privilege.
  • [SAFE]: Includes specialized diagnostic procedures for developers to analyze system alerts via Splunk and Slack (splunk_rw_*, slack_* tools), restricted to authenticated sessions (login --dev).
  • [PROMPT_INJECTION]: Contains an indirect prompt injection surface where the agent processes external data from Slack history and Splunk logs. Ingestion occurs in references/alert-rca.md via slack_get_channel_history and splunk_rw_errors tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 07:05 PM
Security Audit — agent-trust-hub — run-workflow