ln-24-architecture-auditor

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions direct the agent to analyze untrusted external data, which creates a surface for indirect prompt injection where malicious instructions embedded in a repository could attempt to influence the agent's behavior.
  • Ingestion points: The skill (Checklist section 1) instructs the agent to read repository manifests, architecture documents, entrypoints, deployment definitions, and Git history.
  • Boundary markers: There are no explicit delimiters or boundary markers defined to separate the skill's instructions from the untrusted data being audited.
  • Capability inventory: The skill utilizes file system listing, symbol/dependency topology via language servers, Git history inspection, and web research for framework documentation. It does not include file-write or arbitrary command execution capabilities.
  • Sanitization: The instructions do not specify any validation, escaping, or sanitization of the content retrieved from the audited repository.
  • [SAFE]: The skill identifies itself as a read-only auditor and includes explicit constraints such as 'do not modify code or architecture documents during the audit'.
  • [SAFE]: The mention of auditing 'secret ownership' is correctly scoped as an architectural configuration concern rather than an instruction to extract or exfiltrate sensitive credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 04:08 AM
Security Audit — agent-trust-hub — ln-24-architecture-auditor