llm-wiki-adr-memory
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidelines for the agent to recover decision history from PRs, issues, and commits. No malicious instructions, obfuscation, or sensitive data access patterns were detected.
- [NO_CODE]: The skill contains only markdown-based instructions and templates for documentation. It does not execute scripts, download remote resources, or perform shell commands.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data sources such as PR comments and discussion threads. While this presents an attack surface for indirect prompt injection, the skill includes explicit safety gates advising the agent not to invent rationale or treat inferred motivation as fact, which is a recommended defensive practice.
Audit Metadata