persistent-instructions
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of natural language documentation outlining best practices for defining AI agent behavior rules (e.g., in CLAUDE.md or AGENTS.md files).
- [NO_CODE]: There are no scripts, binaries, or automated commands included in the skill. It functions as a reference guide for instruction writing.
- [PROMPT_INJECTION]: No evidence of prompt injection attempts, bypass instructions, or malicious overrides. The guidelines focus on clarity, brevity, and consistency in agent instructions.
Audit Metadata