team-builder
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill reads local markdown files and uses their content as prompts for sub-agents, creating an indirect prompt injection surface.
- Ingestion points: Reads files from
./agents/and~/.claude/agents/. - Boundary markers: None; the agent file content is interpolated directly into the prompt.
- Capability inventory: Spawns sub-agents using the Agent tool to execute task-specific logic.
- Sanitization: None.
- [COMMAND_EXECUTION]: The skill uses the Agent tool to execute personas discovered in markdown files. This is the intended behavior of the orchestrator.
- [SAFE]: No malicious obfuscation, hardcoded credentials, unauthorized network exfiltration, or persistence mechanisms were detected. File access is confined to directories intended for agent personas.
Audit Metadata