task-cost-estimator

本地 Python 命令行工具 task-cost：输入一段任务描述后分析所需能力（推理/编码/写作/研究），匹配 18 个模型并按匹配度排序，给出单次与日成本估算，提供 value、quality、balanced、local 四种模式，用于在开始 AI 任务前挑选性价比最高的模型。

minirr890112-byte

热度 32.21

最近更新 2026-05-11 22:21:47

分类：效率依赖：代码执行应用场景：推荐 / 工作流

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add task-cost-estimator

Metadata

name: task-cost-estimator
description: Before starting any AI task, find the most cost-effective model and estimated cost. 4 modes (value, quality, balanced, local). Task profiling + model matching + cost estimation in one command. Tracks lifetime Bonus.
version: 1.1.0
author: minirr890112-byte
license: MIT
metadata: {"hermes":{"tags":["AI","Cost","Model","Task","Estimator","Budget","Optimization","LLM"],"homepage":"https://github.com/minirr890112-byte/HermesMade"}}

task-cost-estimator

一句话

Before you start any AI task, find the cheapest model that can handle it — and estimate what you'll spend.

核心功能

Task Profiling — 自动分析任务需要什么能力（推理/编码/写作/研究）
Model Matching — 匹配 18 个模型，按匹配度排序
Cost Estimation — 预估单次运行成本、日成本
4 Modes — value（省钱）、quality（质量）、balanced（均衡）、local（本地）

怎么用

task-cost "describe your task here"

示例

$ task-cost "build a REST API with JWT auth in Python"

📋 Task: REST API with JWT auth
   Needs: reasoning=7/10  coding=9/10  writing=3/10
   Est. tokens: ~1,000 in / ~1,000 out

⭐ DeepSeek V4 Flash    $0.0004/run  87% match
   DeepSeek V4 Pro      $0.0027/run  96% match

数据来源

Built from HermesMade's pain point intelligence pipeline. Developers lose money using GPT-4 for simple tasks. This tool gives you the cheapest model that can actually do the job.

Install

pip install git+https://github.com/minirr890112-byte/HermesMade.git#subdirectory=task-cost-estimator

为什么给颗星？

如果你觉得有用，点个 ⭐ 让更多人找到它 → GitHub