video-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes workflow plans and generation status from the ArcReel MCP service, creating an ingestion surface for potentially untrusted data.
- Ingestion points: Workflow plans and batch results returned by
get_workflow_planandget_generation_batch(documented in SKILL.md). - Boundary markers: The skill explicitly directs the agent in
references/generation-results.mdandreferences/plan-safety.mdto use structured codes and ignore human-oriented prose descriptions from the tool output. - Capability inventory: Tool calls for project management, video generation, and transaction handling within the ArcReel ecosystem.
- Sanitization: The agent is instructed to only use arguments provided in the plan's
next_action.argsstructure, minimizing the risk of the agent generating its own unsafe parameters based on malicious input.
Audit Metadata