auto-orchestrator

Pass

Audited by Gen Agent Trust Hub on Apr 5, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill decomposes user-provided requests into instructions for sub-agents without isolation or sanitization.
  • Ingestion points: User requests are captured and split in Step 1 and Step 4 of SKILL.md to generate sub-agent prompts.
  • Boundary markers: Absent. No delimiters or instructions are used to separate user data from the agent's system-level instructions in the Task() prompt field.
  • Capability inventory: The skill utilizes the Task() tool to spawn agents with Bash (command execution) and general-purpose (filesystem write) roles in SKILL.md.
  • Sanitization: Absent. There is no evidence of input validation or escaping before interpolation into sub-agent prompts.
  • [COMMAND_EXECUTION]: Command Execution via Proxy. The skill facilitates the execution of shell commands and filesystem modifications by assigning sub-tasks to Bash and general-purpose agents. This enables a chain where user-controlled input can trigger privileged operations through the Task tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 5, 2026, 08:06 PM
Security Audit — agent-trust-hub — auto-orchestrator