Token导航 LogoToken导航TokenDH.com
开发敏感数据github未标认证来源可访问clear审计通过

project-context-loader项目上下文加载器

Agent Skill

project-context-loader 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

267

周安装

11

GitHub Stars

17

下载量

87
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

3

许可证

MIT

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:project-context-loader(项目上下文加载器)
来源仓库:https://github.com/nguyenthienthanh/aura-frog
仓库路径:skills/project-context-loader
安装命令:
npx skills add https://github.com/nguyenthienthanh/aura-frog --skill project-context-loader
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/nguyenthienthanh/aura-frog --skill project-context-loader

简介

用于处理 GitHub 仓库、Issue 和 Pull Request 信息。

  • 适合批量加载项目元数据和协作历史以增强 Agent 理解。
  • 可解析复杂分支结构和合并请求关系图。
  • 安装命令:npx skills add https://github.com/nguyenthienthanh/aura-frog --skill project-context-loader
  • 注意评估对大型仓库的性能影响和缓存策略。

SKILL.md

Project Context Loader

Priority: HIGH - Use before any workflow or code generation


When to Use

BEFORE:

  • workflow:start or any implementation
  • Any code generation or new files
  • Refactoring or writing tests

SKIP only for: Simple questions (no code), when session-context.toon already loaded


Loading Process

1. Check for Cached Context

# If session context exists and is recent (< 1 hour), use it
if [ -f ".claude/session-context.toon" ]; then
  Read .claude/session-context.toon
  # Skip full scan
fi

2. Generate Session Context (If Missing)

Scan Codebase Patterns

# File naming convention
ls src/components/ | head -5  # PascalCase or kebab-case?

# Import style
Grep: "^import.*from ['\"]@/" --type ts  # Absolute paths?
Grep: "^import.*from ['\"]\.\./" --type ts  # Relative paths?

# Export pattern
Grep: "^export default" --type ts
Grep: "^export (const|function)" --type ts

# Error handling
Grep: "Result<|Either<" --type ts  # Functional?
Grep: "try.*catch" --type ts  # Exception-based?

Generate TOON File

Write to .claude/session-context.toon:

# Session Context - Auto-generated
# Generated: {timestamp}

---

project:
  name: {from package.json}
  stack: {detected}

patterns[6]{type,convention,example}:
  file_naming,{PascalCase|kebab-case},{example}
  imports,{absolute|relative},{example}
  exports,{named|default},{example}
  errors,{result|exceptions},{example}
  testing,{jest|vitest},{example}
  styling,{tailwind|css-modules},{example}

workflow:
  phase: {current}
  feature: {name}
  branch: {git branch}

3. Load Project-Specific Config

Location: .claude/project-contexts/[project]/

FileContainsPriority
project-config.yamlTech stack, integrations, entry pointsAlways load
conventions.mdNaming, code style, patterns, idiomsAlways load
rules.mdProject-specific rulesAlways load
repo-map.mdAnnotated directory tree with purposesLoad on first task
file-registry.yamlKey files: entry points, configs, hub filesLoad when modifying code
architecture.mdModule map, data flow, patterns, dependenciesLoad for architecture decisions

4. Merge Priority

session-context.toon (cached patterns — 12 detections)
  ↓
project-contexts/conventions.md (overrides)
  ↓
project-contexts/repo-map.md + file-registry.yaml + architecture.md (deep context)
  ↓
Aura Frog core rules (defaults)

5. Smart Loading Strategy

Not all context files need loading every time. Use this routing:

loading_strategy[4]{scenario,files_to_load}:
  Simple question (no code),session-context.toon only
  Bug fix / small change,"session-context.toon + conventions.md + file-registry.yaml"
  New feature / refactor,"All 7 files (full context)"
  Architecture decision,"session-context.toon + architecture.md + repo-map.md"

Token budget per scenario:

  • Simple: ~200 tokens
  • Bug fix: ~800 tokens
  • Full context: ~2000 tokens
  • Architecture: ~1000 tokens

Session Context Template

# Session Context
# Generated: 2026-03-20T10:00:00Z
# Valid for: 1 hour (regenerate if stale)

---

project:
  name: my-app
  stack: React,TypeScript,TailwindCSS

patterns[12]{type,convention,example}:
  file_naming,PascalCase,UserProfile.tsx
  imports,absolute @/,import { Button } from '@/components/Button'
  exports,named,export const UserCard = ...
  errors,result,return { ok: true, data }
  testing,vitest,describe('UserCard', () => ...)
  styling,tailwind,className="flex items-center"
  indentation,2-space,2-space
  state_mgmt,zustand,zustand
  api_pattern,react-query,react-query
  components,functional,functional
  env,dotenv-multi,dotenv-multi
  monorepo,none,none

workflow:
  phase: 4
  feature: user-authentication
  branch: feature/user-auth

decisions[2]{id,choice}:
  auth,JWT tokens
  storage,Redis session

Token Efficiency

ApproachTokens
Full repo re-scan~5000-10000
All 7 context files~2000
Session context TOON (12 patterns)~200
Savings vs re-scan95-98%

Commands

Generate Context Manually

bash scripts/context-compress.sh

Clear Context (Force Rescan)

rm .claude/session-context.toon

If Context Missing

⚠️ **Project context not found!**

Generating session-context.toon from codebase scan...
Run `project:init` for full project config.

Related:

  • rules/agent/codebase-consistency.md - Pattern matching rule
  • skills/session-continuation/SKILL.md - Workflow state + handoff
  • scripts/context-compress.sh - Context generator

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

能力 5

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

OpenCode

26.82%
按下载量换算23

Claude Code

26.14%
按下载量换算23

windsurf

19.18%
按下载量换算17

cline

13.55%
按下载量换算12

Codex

7.13%
按下载量换算6

Antigravity

3.9%
按下载量换算3

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。

来源信息

继续浏览同类 Skills