workflow
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and templates do not contain any evidence of malicious activity, unauthorized file access, or credential harvesting. All tools are used in accordance with their intended functionality for project management and software development.
- [PROMPT_INJECTION]: The skill's architecture as an orchestrator involves processing untrusted data (user-defined tasks and codebase content), which creates an inherent attack surface for indirect prompt injection. This risk is characteristic of agent-based orchestration systems and is mitigated by the platform's underlying security controls.
- Ingestion points: User-supplied task descriptions via $ARGUMENTS and the project's source code files accessed during workflow execution.
- Boundary markers: The instructions do not define specific delimiters to isolate potentially untrusted external data from the agents' primary instruction set.
- Capability inventory: Orchestrated sub-agents are granted extensive capabilities, including filesystem modification (Write, Edit) and shell execution (Bash).
- Sanitization: No explicit content sanitization or validation logic is defined within the workflow prompts.
Audit Metadata