跳到正文
TakoAPI
返回技能列表

pinchbench

由 olearycrew 提供

Coding Agents & IDEs
439下载量由 clawskills.sh 提供
0浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Run PinchBench benchmarks to evaluate OpenClaw agent performance across real-world tasks.

描述

该技能的功能

PinchBench measures LLM model performance as the brain of an OpenClaw agent across 23 standardized tasks. Tasks span calendar, email, research, coding, file management, and multi-step workflows. Results are graded automatically and can be submitted to a public leaderboard at pinchbench.com.

Provides 23 reproducible, auto-graded tasks with a public leaderboard, making model comparisons objective rather than anecdotal.

何时使用

  • Comparing two models on email triage and calendar tasks before picking one
  • Running the automated-only suite for a quick performance check before a release
  • Submitting benchmark scores to the PinchBench public leaderboard
  • Testing a specific subset of tasks like market research or PDF summarization
  • Verifying a new OpenClaw setup handles multi-step API workflows correctly

示例工作流

输入

User asks: How well does claude-sonnet-4 handle email and calendar tasks?

智能体
  1. 1

    Register for a PinchBench API token using the --register flag

  2. 2

    Run the benchmark targeting specific tasks: --suite task_01_calendar,task_16_email_triage,task_17_email_search

  3. 3

    Wait for tasks to complete and results to save as JSON in the results/ directory

  4. 4

    Inspect scores with jq to extract per-task grading means

  5. 5

    Results upload automatically to the leaderboard using the registered token

输出

Per-task scores for calendar and email tasks, plus an overall average, visible on the pinchbench.com leaderboard

前置要求

  • PinchBench API token for leaderboard submission (register once via --register flag)

安装

clawhub install olearycrew/pinchbench

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接