Token导航 LogoToken导航TokenDH.com
开发需要联网github未标认证来源可访问许可证需确认审计通过

grace-generate恩典生成

Agent Skill

grace-generate 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

192

周安装

8

GitHub Stars

150

下载量

64
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:grace-generate(恩典生成)
来源仓库:https://github.com/osovv/grace-marketplace
仓库路径:skills/grace-generate
安装命令:
npx skills add https://github.com/osovv/grace-marketplace --skill grace-generate
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/osovv/grace-marketplace --skill grace-generate

简介

grace-generate 用于处理 GitHub 仓库协作相关信息。

  • 支持 Issue、Pull Request 和代码变更的整理与分析。
  • 适用于需要同步团队开发状态的场景。grace-generate 属于开发类 Skill,可作为该场景下的辅助能力补充。
  • 通过 GitHub 安装后,可在 Codex、Claude 等宿主中使用。
  • 使用前需确认仓库权限及是否触发网络请求或文件操作。

SKILL.md

Generate code for the module specified by the user.

Prerequisites

  • docs/development-plan.xml must contain this module's contract
  • docs/knowledge-graph.xml must have this module registered
  • If either is missing, tell the user to run $grace-plan first

Process

Step 1: Load Context

Use the smallest complete context that still preserves contract fidelity.

  1. In standalone generation, read:

- the target module's entry in docs/knowledge-graph.xml - all modules it depends on - the contracts of those dependency modules

  1. In controller-driven multi-agent generation, treat the controller's execution packet as the primary source of truth
  2. If the packet already contains the module contract, graph entry, dependency summaries, write scope, and verification command, do not reread the whole plan and graph again
  3. Read extra local files only when needed to implement the assigned module correctly

Step 2: Generate with GRACE Markup

Generate the code file following this exact structure:

// FILE: path
// VERSION: 1.0.0
// START_MODULE_CONTRACT
//   PURPOSE: from development-plan.xml
//   SCOPE: from development-plan.xml
//   DEPENDS: list
//   LINKS: knowledge graph references
// END_MODULE_CONTRACT
//
// START_MODULE_MAP
//   list all exports with one-line descriptions
// END_MODULE_MAP

// START_CHANGE_SUMMARY
//   LAST_CHANGE: [v1.0.0 - Initial generation from development plan]
// END_CHANGE_SUMMARY

imports

// START_CONTRACT: ComponentOrFunction
//   PURPOSE: ...
//   INPUTS: ...
//   OUTPUTS: ...
// END_CONTRACT: ComponentOrFunction

code with START_BLOCK / END_BLOCK markers every ~500 tokens

Adapt comment syntax to the project language (for example # for Python, // for Go, TypeScript, or Java).

Step 3: Produce Graph Sync Output

Handle graph synchronization differently depending on who owns shared artifacts.

  • Standalone mode: update docs/knowledge-graph.xml directly
  • Controller-driven multi-agent mode: do not edit shared XML directly; return a graph delta proposal containing:

- module ID and file path - actual imports and implied CrossLinks - exports and annotations - any dependency mismatch between contract and implementation

Step 4: Verify

Run verification at module scope unless the caller explicitly asks for more.

  • Run type checking, linting, and module-local tests as appropriate
  • Prefer deterministic local assertions over broad end-to-end checks at this stage
  • Leave wave-level integration checks and full-suite runs to the controller in multi-agent execution

Step 5: Report

Return a concise implementation packet containing:

  1. files changed
  2. module-local verification results
  3. graph sync output or graph delta proposal
  4. open assumptions or integration risks

Granularity Rules

  • Blocks should be ~500 tokens (matching LLM sliding window)
  • Every block name must be unique within the file
  • Block names should describe WHAT the block does, not HOW

- Good: VALIDATE_USER_PERMISSIONS, FETCH_AND_CACHE_DATA - Bad: LOGIC_1, HELPER, MAIN_BLOCK

  • Do not broaden the write scope beyond the assigned module unless the controller or user explicitly revises the plan

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.14%
按下载量换算23

Claude

29.54%
按下载量换算19

Cursor

17.02%
按下载量换算11

Gemini CLI

9.05%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills