itsol-subagent-workflow

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process outputs from delegated subagents, which represents a potential attack surface for indirect prompt injection if a subagent's output contains malicious instructions meant to influence the main agent's behavior.
  • Ingestion points: The main agent ingests data from subagent responses (SKILL.md, 01-planning-and-delegation.md) and processes planning artifacts which could originate from untrusted sources.
  • Boundary markers: The skill establishes strong boundaries through the use of a formal 'Response Contract' and 'Task Packet' system (01-planning-and-delegation.md), requiring subagents to provide evidence-based responses with specific status labels.
  • Capability inventory: The skill possesses the capability to modify files, create Git commits, and execute verification commands (02-review-commit-validation.md).
  • Sanitization: The workflow requires independent review by a different agent and mandatory TDD (Test Driven Development) or documented replacement verification to ensure subagent outputs are valid and safe before integration.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 02:31 PM
Security Audit — agent-trust-hub — itsol-subagent-workflow