eval-recipes-runner

Security scan

AI scan High

VirusTotal scan Unknown

What looks risky

Supply chain or malware 8.5

skill.md

downloads and executes remote payloads

Privacy exfiltration 5.8

skill.md

mentions sensitive credential material

Obfuscation or evasion 5.8

skill.md

contains dynamic execution primitive

Run Microsoft's eval-recipes benchmarks to validate amplihack improvements against baseline agents. Auto-activates when testing improvements, running evals, or benchmarking changes.