to-issues

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [NO_CODE]: The skill contains no executable scripts, binaries, or subprocess calls in its implementation files, consisting only of instructions and metadata.
  • [PROMPT_INJECTION]: The skill processes untrusted external content such as URLs, specifications, and issue bodies to generate tickets, which presents a surface for indirect prompt injection. This risk is effectively mitigated by the skill's mandatory user approval workflow.
  • Ingestion points: SKILL.md Step 2 (Gathering context from roadmaps, plans, PRDs, or external issue URLs).
  • Boundary markers: Not explicitly defined for external content ingestion.
  • Capability inventory: The skill utilizes tool-based access to read repository context and environment variables, and to publish content to external issue trackers (Step 5).
  • Sanitization: The skill includes specific behavioral instructions to the agent to filter out secrets, private keys, and sensitive logs before ticket creation.
  • [SAFE]: The inclusion of a review step (Step 4) and safety constraints regarding sensitive data protection align with security best practices for agent-human interaction.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 12:19 PM
Security Audit — agent-trust-hub — to-issues