delegation-templates

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill implements a framework for subagent delegation that involves prompt interpolation of untrusted data, qualifying as an indirect prompt injection surface.
  • Ingestion points: Data enters the subagent context via the <task>, <context>, <items>, and <session-notes> tags defined in the SKILL.md templates.
  • Boundary markers: The templates use XML-style tags as delimiters to provide structural separation between system instructions and user-provided task descriptions.
  • Capability inventory: The skill is designed to orchestrate agents with various capabilities, including file-system modification (Implementer), investigative research (Researcher), and execution review (Reviewer).
  • Sanitization: No explicit sanitization or validation of the interpolated content is present within the markdown-based templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 05:35 AM
Security Audit — agent-trust-hub — delegation-templates