ccdawn-huawei-nslb-score-loop
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
- [SAFE]: No security issues detected. The skill follows best practices for project-specific automation.
- [COMMAND_EXECUTION]: The skill executes local Python scripts (score_loop_tools.py and sync_project_memory.py) located within the skill's own directory or the user's codex folder. These commands are used for status reporting, epoch management, and memory synchronization, which are standard for its stated purpose.
- [DATA_EXFILTRATION]: The skill facilitates data exchange with an online leaderboard ('online feedback'). This is a legitimate requirement for the Huawei competition and is conducted through defined project tools without evidence of redirection to malicious domains.
- [REMOTE_CODE_EXECUTION]: The skill manages 'workers' that apply patches to local source code (src/Solution.cpp). This execution is isolated to designated workspaces and governed by a mutation matrix that prevents unauthorized or broad code changes.
Audit Metadata