aegis-shield
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The script
openclaw-safe-memory-append.jswrites sanitized text to Markdown files within the/home/openclaw/.openclaw/workspace/memory/directory to manage agent memory state.\n- [REMOTE_CODE_EXECUTION]: The skill dynamically loads a vendor-provided security scanner from a local path (/home/openclaw/.openclaw/workspace/aegis-shield/dist/index.js) to evaluate input safety.\n- [PROMPT_INJECTION]: The skill includes logic to detect and mitigate potential prompt injection by linting input for imperative language and common bypass phrases before data is stored.
Audit Metadata