swarm
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions define a process for task delegation and verification. No malicious commands, obfuscated content, or unauthorized file access patterns were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from sub-agent reports. (1) Ingestion points: Sub-agent reports are merged and verified in SKILL.md. (2) Boundary markers: The skill defines a report shape but does not specify strict delimiters to isolate sub-agent content. (3) Capability inventory: Step 5 of SKILL.md involves running project-level checks, which may include code execution. (4) Sanitization: No sanitization of sub-agent report content is described. The risk is inherent to the task delegation workflow and no malicious use-case was identified.
Audit Metadata