69 lines
28 KiB
JSON
69 lines
28 KiB
JSON
{
|
||
"model": "qwen/qwen3.6-27b",
|
||
"messages": [
|
||
{
|
||
"role": "system",
|
||
"content": "You are \"Connect AI\", a premium agentic AI coding assistant running 100% offline on the user's machine.\nYou are DIRECTLY CONNECTED to the user's local file system and terminal. You MUST use the action tags below to create, edit, delete, read files and run commands. DO NOT just show code — ALWAYS wrap it in the appropriate action tag so it gets executed.\n\nYou have SEVEN powerful agent actions:\n\n━━━ ACTION 1: CREATE NEW FILES ━━━\n<create_file path=\"relative/path/file.ext\">\nfile content here\n</create_file>\n\nExample — user says \"index.html 만들어줘\":\n<create_file path=\"index.html\">\n<!DOCTYPE html>\n<html><head><title>Hello</title></head>\n<body><h1>Hello World</h1></body>\n</html>\n</create_file>\n\n━━━ ACTION 2: EDIT EXISTING FILES ━━━\n<edit_file path=\"relative/path/file.ext\">\n<find>exact text to find</find>\n<replace>replacement text</replace>\n</edit_file>\nYou can have multiple <find>/<replace> pairs inside one <edit_file> block.\n\n━━━ ACTION 3: DELETE FILES ━━━\n<delete_file path=\"relative/path/file.ext\"/>\n\n━━━ ACTION 4: READ FILES ━━━\n<read_file path=\"relative/path/file.ext\"/>\nUse this to read any file in the workspace BEFORE editing it. You will receive the file contents automatically.\n\n━━━ ACTION 5: LIST DIRECTORY ━━━\n<list_files path=\"relative/path/to/dir\"/>\nUse this to see what files exist in a specific subdirectory.\n\n━━━ ACTION 6: RUN TERMINAL COMMANDS ━━━\n<run_command>npm install express</run_command>\n\nExample — user says \"서버 실행해줘\":\n<run_command>node server.js</run_command>\n\n━━━ ACTION 7: READ USER'S SECOND BRAIN (KNOWLEDGE BASE) ━━━\n<read_brain>filename.md</read_brain>\nUse this to READ documents from the user's personal knowledge base.\n\n━━━ ACTION 8: READ WEBSITES & SEARCH INTERNET ━━━\n<read_url>https://example.com</read_url>\nTo search the internet, you MUST use DuckDuckGo by formatting the URL like this:\n<read_url>https://html.duckduckgo.com/html/?q=YOUR+SEARCH+QUERY</read_url>\nUse this forcefully whenever asked for real-time info, news, or whenever requested to \"search\". NEVER say you cannot search.\n\nCRITICAL RULES:\n1. ALWAYS respond in the same language the user uses.\n2. When the user asks to create, edit, delete files or run commands, you MUST use the action tags above. NEVER just show code without action tags.\n3. Outside of action blocks, briefly explain what you did.\n4. For code that is ONLY for explanation (not to be saved), use standard markdown code fences.\n5. Be concise, professional, and helpful.\n6. When editing files, FIRST use <read_file> to read the file, then use <edit_file> with exact matching text.\n7. When a SECOND BRAIN INDEX is available, ALWAYS check it first.\n8. You can use MULTIPLE action tags in a single response.\n9. File paths are RELATIVE to the user's open workspace folder.\n10. The [WORKSPACE INFO] section tells you exactly which folder is open and what files exist. USE this information.\n\n[BACKGROUND CONTEXT]\n\n\n[Currently open file: interfaces.ts]\n```\n복적/**\n * ============================================================\n * Service Interfaces (서비스 인터페이스 정의)\n * \n * 각 서비스(Agent, Brain, FileSystem 등)의 추상화 인터페이스를 정의합니다.\n * 의존성 주입(DI)과 단위 테스트를 위해 필수적입니다.\n * ============================================================\n */\n\nimport * as vscode from 'vscode';\n\n// ─── 에이전트 서비스 인터페이스 ───\nexport interface IAgentService {\n /**\n * LLM에 프롬프트를 보내고 스트리밍 응답을 가져옴\n */\n chat(prompt: string, context: string, model?: string): Promise<AsyncGenerator<string>>;\n\n /**\n * 터미널 명령어 실행 (보안 정책 적용)\n */\n runCommand(command: string): Promise<{ stdout: string; stderr: string }>;\n}\n\n// ─── 파일 시스템 서비스 인터페이스 ───\nexport interface IFileSystemService {\n /**\n * 파일 생성\n */\n createFile(filePath: string, content: string): Promise<void>;\n\n /**\n * 파일 읽기\n */\n readFile(filePath: string): Promise<string>;\n\n /**\n * 파일 수정\n */\n editFile(filePath: string, find: string, replace: string): Promise<void>;\n\n /**\n * 파일 삭제\n */\n deleteFile(filePath: string): Promise<void>;\n\n /**\n * 디렉토리 목록 조회\n */\n listDirectory(dirPath: string): Promise<string[]>;\n}\n\n// ─── 두뇌 서비스 인터페이스 ───\nexport interface IBrainService {\n /**\n * 두뇌 폴더 경로 가져오기\n */\n getBrainDir(): string;\n\n /**\n * 두뇌 폴더가 명시적으로 설정되었는지 확인\n */\n isBrainDirExplicitlySet(): boolean;\n\n /**\n * 두뇌 폴더 내 파일 목록 조회\n */\n getBrainFiles(): Promise<string[]>;\n\n /**\n * 두뇌 파일 내용 읽기\n */\n readBrainFile(fileName: string): Promise<string>;\n}\n\n// ─── 웹뷰 서비스 인터페이스 ───\nexport interface IWebviewService {\n /**\n * 웹뷰에 메시지 전송\n */\n postMessage(message: any): void;\n\n /**\n * 웹뷰에서 메시지 수신 핸들러 등록\n */\n onDidReceiveMessage(callback: (message: any) => void): vscode.Disposable;\n}\n\n// ─── HTTP 서비스 인터페이스 ───\nexport interface IHttpService {\n /**\n * HTTP GET 요청\n */\n get(url: string, options?: any): Promise<any>;\n\n /**\n * HTTP POST 요청\n */\n post(url: string, data: any, options?: any): Promise<any>;\n}\n```\n\n\n[WORKSPACE INFO]\n📂 경로: e:\\Wiki\n\n[프로젝트 파일 구조]\n📁 00_Raw/\n 📁 2026-04-20/\n 📄 ConnectAI_Folder_Analysis_Report.md\n📁 10_Wiki/\n 📁 Topics/\n 📄 2026-04-22_Enemy_AI_Overhaul_Guideline.md\n 📄 2026-04-23_System_Stabilization_Tactical_Boundary_Update.md\n📁 2nd/\n 📁 01_Archive/\n 📁 2026-04-20/\n 📄 _뇌와 팔다리의 분리_ - 관심사의 분리 (Separation of Concerns).md\n 📄 .gitkeep\n 📄 [[Cluster A\n 📄 [[Cluster E\n 📄 01_React-Foundation_Mental-Model.md\n 📄 01_WebWorker-performance-optimization.md\n 📄 02_React-Hooks_Advanced-Patterns.md\n 📄 02_StateManagement-single-source-of-truth.md\n 📄 03_Architecture-design-principle.md\n 📄 03_React-Architecture_Component-Patterns.md\n 📄 04_execution-environment-management.md\n 📄 04_React-Performance_Optimization.md\n 📄 05_React-State-Management_Strategy.md\n 📄 05_simulation-design-principles.md\n 📄 06_React-Clean-Code_Etiquette.md\n 📄 07_React-Testing_Reliability.md\n 📄 08_React-API_Communication-Patterns.md\n 📄 09_Styling-Architecture_Governance.md\n 📄 10_Modern-Environment_Ecosystem.md\n 📄 11_Reliability_Safety-First.md\n 📄 12_Deployment_Final-Gate.md\n 📄 13_TypeScript_Type-Safety.md\n 📄 14_Accessibility_Inclusivity.md\n 📄 15_Collaboration_Governance.md\n 📄 2026-04-15.md\n 📄 20k skinned instances demo.md\n 📄 3D Gaussian Splatting (3DGS).md\n 📄 3D Web-based HMI.md\n 📄 3D_Gaussian_Splatting.md\n 📄 3D_Web_HMI.md\n 📄 A_B-Testing-Platforms.md\n 📄 ABA.md\n 📄 ABA(Applied Behavior Analysis).md\n 📄 Abstract Syntax Tree (AST).md\n 📄 Abstract-Syntax-Tree-Transformation.md\n 📄 Abstract-Syntax-Tree-Traversal.md\n 📄 Accessibility (A11y).md\n 📄 Accessibility-Compliance-Audit.md\n 📄 Accessibility-Compliance-WCAG.md\n 📄 Accessibility.md\n 📄 ACL_Prevention.md\n 📄 ACL-Injury-Prevention-Protocols.md\n 📄 Adaptive Compute (적응형 계산량 조절).md\n 📄 Adaptive_Learning.md\n 📄 Adaptive-Learning-Systems.md\n 📄 Addiction Neuroscience.md\n 📄 Addiction_Neuroscience.md\n 📄 Additive-Type-Logic.md\n 📄 Advanced-Design-Patterns-in-TypeScript.md\n 📄 Advanced-Interface-Design.md\n 📄 Adversarial Attack (적대적 공격).md\n 📄 Adversarial Code Stylometry.md\n 📄 Aerospace Flight Simulation.md\n 📄 Affective Computing.md\n 📄 Affective User Interfaces (AUI).md\n 📄 agargaro의 오픈 소스 라이브러리.md\n 📄 Agency and Player Autonomy.md\n 📄 Agency-in-Game-Design.md\n 📄 Agency-Narrative Integration.md\n 📄 Agent Communication Protocol (에이전트 통신 규약).md\n 📄 Agent-Based Modeling (ABM).md\n 📄 Agent-Based Modeling.md\n 📄 Agent-Based-Modeling.md\n 📄 Agile-Software-Development-Life-Cycle.md\n 📄 Agile-UX-Integration.md\n 📄 AI Connect LLM Tool.md\n 📄 AI Safety (AI 안전).md\n 📄 AI 거버넌스 정책(AI Usage Policy).md\n 📄 AI 생성 코드 검증(AI Code Assurance).md\n 📄 AI 에이전트 (AI Agent).md\n 📄 AI 코드 리뷰 및 보안 취약점 점검(DevSecOps).md\n 📄 AI 코드 리뷰.md\n 📄 AI-Driven Narrative Systems.md\n 📄 AI.md\n 📄 AI와 기계에게 검열 맡기기_ - 정적 분석 툴 (ESLint Prettier)).md\n 📄 AI와 기계에게 검열 맡기기_ - 정적 분석 툴 (ESLint & Prettier)).md\n 📄 Albion Online (Full Loot_Player-Driven Production).md\n 📄 Albion Online (Full LootPlayer-Driven Production).md\n 📄 Algebraic-Data-Types-in-TypeScript.md\n 📄 Algebraic-Data-Types.md\n 📄 Algorithmic Bias in Art.md\n 📄 Algorithmic Decision Making.md\n 📄 Algorithmic Game Theory.md\n 📄 Algorithmic Governance.md\n 📄 Algorithmic Mechanism Design.md\n 📄 Algorithmic Rhetoric.md\n 📄 Algorithmic-Biology.md\n 📄 Algorithmic-Game-Theory.md\n 📄 Algorithmic-Governance.md\n 📄 Allocation Timeline.md\n 📄 Allocation Timeline(할당 타임라인).md\n 📄 Alpha Blending.md\n 📄 AlphaGo (Monte Carlo Tree Search RL)] [Autonomous Driving Simulation] [Robotic Manipulation.md\n 📄 AlphaGo (\n\n\n[CRITICAL: SECOND BRAIN INDEX — User's Personal Knowledge Base (2851 documents)]\nThe user has synced a personal knowledge repository. Below is the TABLE OF CONTENTS.\n(⚠️ 메모리 폭발 방지를 위해 상위 200개 파일의 목차만 표시됩니다.)\nIf the user's query is even slightly related to any topics in this index, YOU MUST FIRST READ the relevant document BEFORE answering.\nTo read the actual content of any document, use EXACTLY this syntax: <read_brain>filename_or_path</read_brain>\nYou can call <read_brain> multiple times. ALWAYS READ THE FULL DOCUMENT BEFORE ANSWERING.\n\n**IMPORTANT: When your answer uses knowledge from the Second Brain, you MUST end your response with a \"📚 출처\" section listing the file(s) you referenced. Example:\n📚 출처: MrBeast_분석.md, 마케팅_전략.md**\n\n 📄 00_Raw\\2026-04-20\\API 응답 모델링 및 상태 머신(State Machine) 설계.md → \"\"\n 📄 00_Raw\\2026-04-20\\Diffusion-Models.md → \"\"\n 📄 00_Raw\\2026-04-20\\Looking-Glass-Studios.md → \"\"\n 📄 00_Raw\\2026-04-20\\SAST.md → \"\"\n 📄 01_Frontend_Mastery\\React_Clean_Code_Best_Practices.md → \"---\r\"\n 📄 01_Frontend_Mastery\\React_Hooks_Deep_Dive.md → \"---\r\"\n 📄 01_Frontend_Mastery\\React_Mental_Model.md → \"---\r\"\n 📄 01_Frontend_Mastery\\React_Performance_Optimization.md → \"---\r\"\n 📄 01_Frontend_Mastery\\React_State_Management_Strategy.md → \"---\r\"\n 📄 01_Frontend_Mastery\\React_Testing_Strategy.md → \"---\r\"\n 📄 01_Frontend_Mastery\\TypeScript 컴파일러의 정적 타입 시스템.md → \"\"\n 📄 01_Frontend_Mastery\\TypeScript_Type_Safety.md → \"---\r\"\n 📄 01_Frontend_Mastery\\WebWorker_Performance.md → \"---\r\"\n 📄 02_Architecture_Principles\\API_Communication_Patterns.md → \"---\r\"\n 📄 02_Architecture_Principles\\Component_Design_Patterns.md → \"---\r\"\n 📄 02_Architecture_Principles\\Separation_of_Concerns.md → \"---\r\"\n 📄 02_Architecture_Principles\\Single_Source_of_Truth.md → \"---\r\"\n 📄 02_Architecture_Principles\\Systemic_Simulation_Principles.md → \"---\r\"\n 📄 03_DevOps_Environment\\Deployment_Final_Gate.md → \"---\r\"\n 📄 03_DevOps_Environment\\Developer Produ"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "안녕"
|
||
},
|
||
{
|
||
"role": "assistant",
|
||
"content": "\n\n안녕하세요! 👋 오늘 어떤 작업이 필요하신가요?"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "지금 E:\\Wiki\\ConnectAI 이폴더 분석해줘. 설계적으로, 기능적으로, 사용자 입장에서 개선이 필요하거나 문제가 있거나 최적화가 안된 부분이 있거나 기타 등등.."
|
||
},
|
||
{
|
||
"role": "assistant",
|
||
"content": "\n\n먼저 ConnectAI 폴더의 전체 구조를 파악하겠습니다.\n\n<list_files path=\"ConnectAI\"/>"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "[시스템: list_files 결과]\n디렉토리: ConnectAI/\n📁 assets/\n📄 brain-viz.html\n📄 connect-ai-lab-2.2.15.vsix\n📄 diff.txt\n📄 diff_utf8.txt\n📄 g1nation-2.2.15.vsix\n📄 g1nation-final-v2.vsix\n📄 g1nation-final-v3.vsix\n📄 g1nation-final.vsix\n📄 g1nation-latest.vsix\n📄 g1nation-new-build.vsix\n📄 LICENSE\n📄 package-lock.json\n📄 package.json\n📄 README.md\n📁 src/\n📄 system_schema.json\n📄 tsconfig.json"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "지금 E:\\Wiki\\ConnectAI 이폴더 분석해줘. 설계적으로, 기능적으로, 사용자 입장에서 개선이 필요하거나 문제가 있거나 최적화가 안된 부분이 있거나 기타 등등.."
|
||
},
|
||
{
|
||
"role": "assistant",
|
||
"content": "\n\n먼저 핵심 파일들을 모두 읽어서 프로젝트 전체를 파악하겠습니다.\n\n<read_file path=\"ConnectAI/package.json\"/>\n<read_file path=\"ConnectAI/tsconfig.json\"/>\n<read_file path=\"ConnectAI/README.md\"/>\n<read_file path=\"ConnectAI/system_schema.json\"/>\n<list_files path=\"ConnectAI/src\"/>"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "[시스템: read_file 결과]\n파일: ConnectAI/package.json\n```\n{\n \"name\": \"g1nation\",\n \"displayName\": \"G1nation\",\n \"description\": \"100% local AI coding agent for VS Code. Create files, edit code, run commands, and work offline with Ollama or LM Studio.\",\n \"version\": \"2.2.15\",\n \"publisher\": \"connectailab\",\n \"license\": \"MIT\",\n \"icon\": \"assets/icon.png\",\n \"repository\": {\n \"type\": \"git\",\n \"url\": \"https://github.com/wonseokjung/connect-ai\"\n },\n \"engines\": {\n \"vscode\": \"^1.80.0\"\n },\n \"categories\": [\n \"Machine Learning\",\n \"Programming Languages\",\n \"Chat\"\n ],\n \"keywords\": [\n \"ai\",\n \"local\",\n \"ollama\",\n \"gemma\",\n \"llama\",\n \"deepseek\",\n \"offline\",\n \"agent\",\n \"code-generation\",\n \"g1nation\",\n \"copilot\"\n ],\n \"activationEvents\": [\n \"*\"\n ],\n \"main\": \"./out/extension.js\",\n \"contributes\": {\n \"commands\": [\n {\n \"command\": \"g1nation.newChat\",\n \"title\": \"G1nation: New Chat\",\n \"icon\": \"$(add)\"\n },\n {\n \"command\": \"g1nation.exportChat\",\n \"title\": \"G1nation: Export Chat as Markdown\"\n },\n {\n \"command\": \"g1nation.explainSelection\",\n \"title\": \"G1nation: Explain Selected Code\"\n },\n {\n \"command\": \"g1nation.focusChat\",\n \"title\": \"G1nation: Focus Chat Input\"\n },\n {\n \"command\": \"g1nation.showBrainNetwork\",\n \"title\": \"G1nation: Show Brain Topology\"\n }\n ],\n \"keybindings\": [\n {\n \"command\": \"g1nation.focusChat\",\n \"key\": \"cmd+l\",\n \"mac\": \"cmd+l\"\n }\n ],\n \"menus\": {\n \"editor/context\": [\n {\n \"command\": \"g1nation.explainSelection\",\n \"when\": \"editorHasSelection\",\n \"group\": \"1_modification\"\n }\n ]\n },\n \"viewsContainers\": {\n \"activitybar\": [\n {\n \"id\": \"g1nation-sidebar\",\n \"title\": \"G1nation\",\n \"icon\": \"$(hubot)\"\n }\n ]\n },\n \"views\": {\n \"g1nation-sidebar\": [\n {\n \"type\": \"webview\",\n \"id\": \"g1nation-v2-view\",\n \"name\": \"Chat\"\n }\n ]\n },\n \"configuration\": {\n \"title\": \"G1nation\",\n \"properties\": {\n \"g1nation.ollamaUrl\": {\n \"type\": \"string\",\n \"default\": \"http://127.0.0.1:11434\",\n \"description\": \"Base URL for Ollama or LM Studio. Default: http://127.0.0.1:11434\"\n },\n \"g1nation.defaultModel\": {\n \"type\": \"string\",\n \"default\": \"gemma4:e2b\",\n \"description\": \"Default model name to use for chat requests.\"\n },\n \"g1nation.requestTimeout\": {\n \"type\": \"number\",\n \"default\": 300,\n \"description\": \"Request timeout in seconds. Default: 300\"\n },\n \"g1nation.localBrainPath\": {\n \"type\": \"string\",\n \"default\": \"\",\n \"description\": \"Folder path for your local Second Brain knowledge base. Leave empty to use the default folder.\"\n },\n \"g1nation.secondBrainRepo\": {\n \"type\": \"string\",\n \"default\": \"\",\n \"description\": \"Optional GitHub repository URL used for Second Brain sync.\"\n },\n \"g1nation.autoPushBrain\": {\n \"type\": \"boolean\",\n \"default\": false,\n \"description\": \"Automatically commit and push Second Brain changes after updates.\"\n }\n }\n }\n },\n \"scripts\": {\n \"vscode:prepublish\": \"npm run compile\",\n \"compile\": \"esbuild src/extension.ts --bundle --platform=node --external:vscode --outfile=out/extension.js\",\n \"watch\": \"tsc -watch -p ./\",\n \"pretest\": \"npm run compile\"\n },\n \"devDependencies\": {\n \"@types/node\": \"18.x\",\n \"@types/vscode\": \"^1.80.0\",\n \"@vercel/ncc\": \"^0.38.4\",\n \"esbuild\": \"^0.28.0\",\n \"typescript\": \"^5.1.3\"\n },\n \"dependencies\": {\n \"axios\": \"^1.15.0\",\n \"jsdom\": \"^29.0.2\"\n }\n}\n\n```"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "[시스템: read_file 결과]\n파일: ConnectAI/tsconfig.json\n```\n{\r\n \"compilerOptions\": {\r\n \"module\": \"commonjs\",\r\n \"target\": \"ES2022\",\r\n \"outDir\": \"out\",\r\n \"lib\": [\"ES2022\"],\r\n \"sourceMap\": true,\r\n \"strict\": true\r\n },\r\n \"exclude\": [\"node_modules\", \".vscode-test\"]\r\n}\r\n\n```"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "[시스템: read_file 결과]\n파일: ConnectAI/README.md\n```\n<p align=\"center\">\r\n <img src=\"assets/icon.png\" width=\"120\" alt=\"Connect AI Logo\" />\r\n</p>\r\n\r\n<h1 align=\"center\">Connect AI v2 (P-Reinforce)</h1>\r\n\r\n<p align=\"center\">\r\n <strong>100% Local · 100% Offline · Autonomous Knowledge Engine</strong><br/>\r\n VS Code / Cursor 확장 프로그램으로, 당신의 낡은 IDE를 최상위 에이전트 대학(A.U)의 심장으로 진화시킵니다.\r\n</p>\r\n\r\n<p align=\"center\">\r\n <img src=\"https://img.shields.io/badge/version-2.1.30-blue\" alt=\"version\" />\r\n <img src=\"https://img.shields.io/badge/license-MIT-green\" alt=\"license\" />\r\n <img src=\"https://img.shields.io/badge/integration-Agent_University-purple\" alt=\"integration\" />\r\n <img src=\"https://img.shields.io/badge/engine-Ollama%20%7C%20LM%20Studio-orange\" alt=\"engine\" />\r\n</p>\r\n\r\n---\r\n\r\n## 🌟 Overview: The P-Reinforce Architecture\r\n\r\nConnect AI v2.1.30은 단순한 코딩 에이전트를 넘어섭니다. **P-Reinforce 아키텍처**를 기반으로 설계된 이 에이전트는 사용자의 모든 정보와 지시를 받아들여 **스스로 의미를 분석하고, 폴더를 생성하고, 마크다운 위키 파일로 정리하여 클라우드에 자동 백업**하는 자율 지식 정원사(Autonomous Gardener)입니다.\r\n\r\n---\r\n\r\n## ⚡ Core Features\r\n\r\n### 1. 🧠 Agent University (A.U) 완벽 연동\r\nAgent University 웹 플랫폼과 실시간으로 통신합니다. \r\n웹에서 버튼 한 번 누르는 즉시, 로컬 VS Code의 `4825` 포트를 통해 프리미엄 브레인 팩(Premium Brain Pack) 지식이 로컬 인공지능 뇌(`~/.connect-ai-brain`)에 자동 주입되어 신경망을 확장합니다.\r\n\r\n### 2. 📂 자율 지식 구조화 (Zero-Interaction Styling)\r\n유저가 던져주는 원시 데이터(Raw Data)를 에이전트가 스스로 판단해 `10_Wiki`, `00_Raw`, `🚀 Skills` 와 같은 완벽한 P-Reinforce 템플릿 규격의 Markdown 파일로 분할-조립하여 저장합니다.\r\n\r\n### 3. ☁️ 클라우드 동기화 (Auto-Git Sync 100%)\r\n로컬 PC에서 파일 생성이 일어나는 순간, 에이전트가 스스로 GitHub 저장소에 `git add`, `commit`, `push`를 수행합니다. \r\n마스터는 이제 지루한 푸시 커맨드를 입력할 필요가 없습니다.\r\n\r\n### 4. 🔗 설치형 모델 자동 감지 (Dynamic Model Detection)\r\nOllama 또는 LM Studio에 설치된 모델을 내부 API(`v1/models`)를 호출하여 자동 감지하고, UI의 스위치 보드(드롭다운)에 연결합니다. 어떤 모델을 쓸지 번거롭게 입력하지 마십시오.\r\n\r\n---\r\n\r\n## ⚒️ Agent Capabilities (에이전트 권한)\r\n\r\n로컬 머신의 파일 시스템과 터미널에 대한 통제권을 인공지능에게 부여합니다. (100% 안전한 권한 승인 기반)\r\n\r\n| Action | Description |\r\n|:--|:--|\r\n| **📄 Create Files** | 새로운 파일과 폴더를 생성합니다 |\r\n| **✏️ Edit Files** | 기존 파일 내의 코드를 수정합니다 |\r\n| **🗑️ Delete Files** | 불필요한 파일을 즉각 파쇄합니다 |\r\n| **📖 Read Files** | 마스터의 프로젝트 파일을 읽어 맥락을 파악합니다 |\r\n| **📂 Browse Directories** | 디렉토리 구조를 분석합니다 |\r\n| **🖥️ Run Commands** | `npm run build`, `git push` 등 터미널 명령을 수행합니다 |\r\n\r\n---\r\n\r\n## 📥 Installation (설치 방법)\r\n\r\n### A.U 멤버십 유저 (Recommended)\r\n1. 상단 탭의 [Releases](https://github.com/wonseokjung/connect-ai/releases) 메뉴로 진입.\r\n2. 최신 `v2.1.30.vsix` 파일을 다운로드.\r\n3. VS Code 에서 `Cmd+Shift+P` → **Extensions: Install from VSIX** → 다운받은 파일 선택\r\n\r\n### 개발자 빌드 (Build from Source)\r\n```bash\r\ngit clone https://github.com/wonseokjung/connect-ai.git\r\ncd connect-ai\r\nnpm install\r\nnpm run compile\r\nnpx vsce package\r\n```\r\n\r\n---\r\n\r\n## ⚙️ Engine Setup (엔진 설정 방법)\r\n\r\n### ✅ LM Studio (Apple Silicon, Windows) - 권장\r\n1. [lmstudio.ai](https://lmstudio.ai/) 에서 설치\r\n2. Gemma 3, Llama 3 또는 Qwen Coder 등 원하는 모델 로드\r\n3. **Developer 탭(좌측 `<>` 메뉴)** 진입 후 **Start Server** 클릭\r\n4. Connect AI의 ⚙️ 채팅방 설정에서 엔진을 \"LM Studio\"로 선택 (자동 모델 인덱싱 완료)\r\n\r\n### ✅ Ollama (Mac, Linux)\r\n```bash\r\nbrew install ollama\r\nollama pull gemma3 # 원하는 모델 풀링\r\n```\r\nConnect AI에서 설정만 \"Ollama\"로 바꿔주시면 끝납니다.\r\n\r\n---\r\n\r\n## 🔒 Privacy (완벽한 보안)\r\n\r\n- **Zero Cloud API:** 당신의 코드는 외부 클라우드 통신망을 타지 않습니다.\r\n- **Zero Telemetry:** 모든 연산력은 100% Local Inference 환경에서 이루어집니다.\r\n- 기업 보안 등급에 준하는 극강의 밀폐형 로컬 지식망 생성을 보장합니다.\r\n\r\n---\r\n\r\n<p align=\"center\">\r\n <strong>Built for Antigravity & Agent University</strong><br/>\r\n Designed by <a href=\"https://github.com/wonseokjung\">Jay</a> × Connect AI Architect\r\n</p>\r\n\n```"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "[시스템: read_file 결과]\n파일: ConnectAI/system_schema.json\n```\n{\r\n \"schema_name\": \"ConnectAI_AgentSystemSchema\",\r\n \"version\": \"1.0.0\",\r\n \"description\": \"A comprehensive JSON schema defining the operational state, capabilities, and configuration of the Connect AI agent.\",\r\n \"agent_identity\": {\r\n \"name\": \"Connect AI\",\r\n \"role\": \"Agentic AI Coding Assistant\",\r\n \"core_principle\": \"100% Local · 100% Offline · 100% Free\"\r\n },\r\n \"environment\": {\r\n \"workspace_path\": \"/Users/jay/로컬테스트/local-ai-coder\",\r\n \"file_system_access\": \"Direct, offline access to the local file system via agent actions.\",\r\n \"context_source\": \"Local workspace files and optional GitHub knowledge base.\"\r\n },\r\n \"agent_capabilities\": [\r\n {\r\n \"action\": \"CREATE_FILE\",\r\n \"description\": \"Generates new files and directories in the workspace.\"\r\n },\r\n {\r\n \"action\": \"EDIT_FILE\",\r\n \"description\": \"Finds and replaces specific code in existing files.\"\r\n },\r\n {\r\n \"action\": \"DELETE_FILE\",\r\n \"description\": \"Removes files and folders.\"\r\n },\r\n {\r\n \"action\": \"READ_FILE\",\r\n \"description\": \"Reads the content of any file in the workspace to understand context.\"\r\n },\r\n {\r\n \"action\": \"LIST_DIRECTORY\",\r\n \"description\": \"Lists contents of a specific subdirectory.\"\r\n },\r\n {\r\n \"action\": \"RUN_TERMINAL_COMMAND\",\r\n \"description\": \"Executes CLI commands (e.g., npm install, node).\"\r\n },\r\n {\r\n \"action\": \"READ_BRAIN\",\r\n \"description\": \"Queries the personal knowledge base (Second Brain) from a linked repository.\"\r\n // Note: Reading URLs is also available as an action if needed.\r\n }\r\n ],\r\n \"configuration\": {\r\n \"engine_options\": {\r\n \"default_model\": \"gemma4:e2b\",\r\n \"ollama_url\": \"http://127.0.0.1:11434\"\r\n },\r\n \"operational_settings\": {\r\n \"temperature\": 0.7,\r\n \"top_p\": 0.9,\r\n \"top_k\": 40,\r\n \"request_timeout_seconds\": 300,\r\n \"max_context_files\": 200\r\n },\r\n \"knowledge_base\": {\r\n \"second_brain_repo_url\": \"\",\r\n \"data_location\": \"~/.connect-ai-brain/\"\r\n }\r\n },\r\n \"workspace_snapshot\": {\r\n \"files_present\": [\r\n \"assets/icon.png\",\r\n \"src/*.ts\",\r\n \"fix_final.py\",\r\n \"README.md\",\r\n \"package.json\",\r\n \"tsconfig.json\"\r\n ],\r\n \"file_count\": 48\r\n }\r\n}\n```"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "지금 E:\\Wiki\\ConnectAI 이폴더 분석해줘. 설계적으로, 기능적으로, 사용자 입장에서 개선이 필요하거나 문제가 있거나 최적화가 안된 부분이 있거나 기타 등등.."
|
||
},
|
||
{
|
||
"role": "assistant",
|
||
"content": "\n\n<list_files path=\"ConnectAI/src\"/>"
|
||
},
|
||
{
|
||
"role": "user",
|
||
"content": "지금 E:\\Wiki\\ConnectAI 이폴더 분석해줘. 설계적으로, 기능적으로, 사용자 입장에서 개선이 필요하거나 문제가 있거나 최적화가 안된 부분이 있거나 기타 등등.."
|
||
}
|
||
],
|
||
"stream": true,
|
||
"max_tokens": 4096,
|
||
"temperature": 0.8,
|
||
"top_p": 0.9
|
||
} |