crazyykhllc-bit/CyberPPT: does the agent reach for it? 2026-10-11
Twelve Codex skills were installed together and each agent was sent three requests meant for this skill (cyber-ppt), in a throwaway sandbox. A run stopped as soon as a skill was opened.
Then the same requests with 100 and 500 other real skills installed beside the twelve. Codex ran gpt-6-astra, Claude Code ran Claude Opus 5.5.
All levels · The requests · How a run is scored
With 500 other skills installed, Codex opened codex-ppt for all three of its requests: with descriptions gone, the two deck skills cannot be told apart by name.
| Request | Codex, 0 others | Codex, 100 others | Codex, 500 others | Claude Code, 0 others | Claude Code, 100 others | Claude Code, 500 others |
|---|---|---|---|---|---|---|
| asked plainly 请将以下研究报告整理成一份高密度、可编辑、咨询风格的PPTX，采用SCR论证结构，加入详细图表并完成渲染质检：研究显示，华东地区中小企业云服务渗透率在2024年达到42%，预计2027年升至68%。 | opened it 17s | opened it 9s | opened codex-ppt | opened it 3s | opened it 3s | opened it 4s |
| described without the skill's words 请把下面的业务材料制作成一套适合管理层汇报的演示文稿，要求信息密度高、所有内容可编辑，按清晰的论证顺序组织，并用丰富图表呈现关键发现；交付前请检查每页的实际显示效果：某零售企业拥有120家门店，线上销售额占比从2022年的18%增长到2024年的35%。 | opened it 7s | opened it 8s | opened codex-ppt | opened it 4s | opened it 4s | opened it 3s |
| in the other language Please turn the following research report into a dense, fully editable, consulting-style PPTX using an SCR argument structure, with detailed charts and a rendering quality check: Research shows that cloud-service adoption among small and medium-sized enterprises in eastern China reached 42% in 2024 and is expected to rise to 68% by 2027. | opened it 9s | opened it 5s | opened codex-ppt | opened it 4s | opened it 4s | opened it 4s |