super-self-improving

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The README and SKILL files reference a Python script named super_self_improving.py and a command-line tool super-self-improving which are not present in the analyzed file set.
  • [SAFE]: No instances of hardcoded credentials, malicious network operations, or unauthorized file system access were identified. The use of a hidden directory in the home folder (~/.super-self-improving/) for storage is consistent with the skill's documented purpose for persistent memory management.
  • [PROMPT_INJECTION]: The skill describes an architecture for learning from 'explicit' and 'implicit' feedback. This creates a surface for indirect prompt injection where adversarial data ingested as feedback could influence the agent's meta-learning logic. However, no evidence of malicious instructions was found in the static files, and this behavior is core to the skill's intended functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 01:35 PM