菜单
TokenSpeed benchmark planning skill for LLM inference experiments.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate real Word, Excel, PowerPoint, PDF, and Markdown files from a structured JSON specification and save them in the current conversation output directory.
Default browser tool for general web automation without explicit authorization required.
Agent rules and review heuristics adapted for OctoAgent coding agents.
HTML deck and slide template selection skill for reports and courseware.
Content experiment skill for calibrated publishing workflows.
Technical diagram generation skill for architecture and workflow visuals.
| name | tokenspeed-benchmark |
| description | TokenSpeed benchmark planning skill for LLM inference experiments. |
| license | MIT |
Define model, hardware, prompts, batch/concurrency, latency, throughput, correctness checks, and baseline comparison before installing engines.
get_plugin_command.integrated_workflow_run when a workflow ID is available.