parallel-agents
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- PROMPT_INJECTION (SAFE): No malicious overrides, bypass instructions, or jailbreak attempts were detected. The skill purely describes structural orchestration patterns.
- DATA_EXFILTRATION (SAFE): No hardcoded secrets, sensitive file paths, or network-based exfiltration patterns (curl, wget) were found.
- REMOTE_CODE_EXECUTION (SAFE): No external dependencies, package installations, or remote script execution patterns were identified.
- COMMAND_EXECUTION (SAFE): The skill is restricted to read-only operations via the Read, Glob, and Grep tools.
- INDIRECT_PROMPT_INJECTION (SAFE): The skill describes patterns for ingesting data but does not perform unsafe interpolation or execution.
- Ingestion points: Uses standard Read/Glob/Grep for codebase analysis.
- Boundary markers: N/A as this is a documentation template.
- Capability inventory: Read-only file access.
- Sanitization: N/A for structural guides.
Audit Metadata