competition-zip-archive

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to record and substitute recovered internal keys and the final flag verbatim into commands and evidence, which requires the LLM to output secret values directly.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). It explicitly instructs running PowerShell with "-ExecutionPolicy Bypass", which directs the agent to bypass a Windows security mechanism (execution policy), so it pushes the agent to circumvent system security.