mirror of
https://github.com/MervinPraison/PraisonAI.git
synced 2026-11-07 12:47:55 -10:00
- Replace eval(e) in unified_server.py with AST-based safe math evaluator - Replace exec(script_content) in run_with_langtrace.py with subprocess.run() - Add nh3.clean() output sanitization in api.py to prevent XSS from markdown links - Replace eval() in basic-tools.ts and custom-tool.ts with regex allowlist + Function constructor - Replace eval() in simplified-api-test.ts and comprehensive-live.test.ts with safe evaluator All changes are in example files and dev helper scripts only. No SDK core code modified. Unit tests: 4341 passed, 0 new failures.