Remote Code Execution (RCE) on Agentic Applications
20Resource Fetch Success RateWindsurf
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| WindsurfModel=SWE-1.5, Version=1.13.52026.07 | 20 | 13 | 65 | |
| Cursor CLIModel=Grok Code, Version=2.2.202026.07 | 17 | 6 | 30 | |
| ClineModel=KAT-Coder-Pro V1, Version=3.66.02026.07 | 11 | 9 | 45 | |
| Gemini CLIModel=Gemini 2.5, Version=0.24.52026.07 | 10 | 4 | 20 | |
| OpenClawModel=GPT-5.4 Codex, Version=v2026.3.23-22026.07 | 10 | 4 | 40 | |
| OpenClawModel=Sonnet 4.6, Version=v2026.3.23-22026.07 | 10 | 10 | 100 | |
| ZeroClawModel=Sonnet 4.6, Version=0.6.82026.07 | 10 | 10 | 100 | |
| NanoClawModel=Sonnet 4.6, Version=1.2.522026.07 | 10 | 10 | 100 | |
| OpenClawModel=Opus 4.6, Version=v2026.3.23-22026.07 | 8 | 10 | 80 | |
| CursorModel=Sonnet 4.5, Version=2.4.222026.07 | 7 | 5 | 25 | |
| Copilot ChatModel=GPT-4.1, Version=0.36.22026.07 | 7 | 7 | 35 |