formal-verify
Warn
Audited by Socket on Apr 2, 2026
1 alert found:
SecuritySecurityscripts/verify-behavioral.py
MEDIUMSecurityMEDIUM
scripts/verify-behavioral.py
This code fragment is not overtly malicious (no hardcoded secrets, no network/subprocess actions). However, it is a high-impact arbitrary code execution harness: it dynamically loads and execs every Python file found in a CLI-specified directory and then runs module.verify() with unvalidated facts. If an attacker can tamper with specs_dir/spec files or influence CLI arguments, the runner can execute attacker-controlled code with its full privileges. Exception/traceback details and spec-produced violation content may be exposed in logs/output.
Confidence: 72%Severity: 78%
Audit Metadata