Glossary: AI Agent

28 terms

GitHub Agent HQ / AgentHQ / GitHub Multi-Agent Orchestration

GitHub's orchestration platform for launching Copilot Workspace, Codex, and Claude Code in parallel to automate large-scale repo work.

AI Agent3

Agent Loop

Agent Loop / エージェントループ / 推論ループ

The cycle an AI agent repeats — reason, call a tool, receive results, re-reason — until a goal or stop condition is reached.

AI Agent3

Agent View

Agent View / Claude Code Agent View / エージェントビュー

Anthropic's May 2026 dashboard for Claude Code — 'claude agents' manages parallel sessions with automatic worktree isolation in one view.

AI Agent3

AI Agent

AI Agent / AIエージェント / 自律型AIエージェント

An AI system that autonomously reasons, plans, and calls tools over multiple steps to complete a goal.

AI Agent3

Aider

Aider AI / aider-chat / Aider CLI

Open-source terminal AI pair programmer deeply integrated with Git, auto-committing code changes from natural language instructions.

AI Agent3

Background Agent

Background Agent / バックグラウンドエージェント / 非同期エージェント

Agent mode where tasks run server-side without the user terminal — Claude Code agents keep working after you close and notify on completion.

AI Agent3

Browser Use

Browser Use Agent / ブラウザ操作エージェント

Open-source library letting AI agents control web browsers via Playwright so LLMs can autonomously scrape, fill forms, and click UIs.

AI Agent3

Claude Code

Claude Code CLI / Anthropic Claude Code / クロードコード

Anthropic's terminal-native AI coding agent with multi-session parallelism managed via the May 2026 Agent View dashboard.

AI Agent3

Cline (VS Code Extension)

Cline VS Code / Cline拡張機能 / Claude Dev

Open-source VS Code AI coding agent supporting multiple LLMs with autonomous file editing, terminal execution, and browser control.

AI Agent3

Computer Use (Anthropic)

Anthropic Computer Use / コンピュータ操作AI / Claude Computer Use

Anthropic's Claude capability for autonomous desktop control via screenshots and mouse/keyboard, enabling agents to execute any GUI task.

AI Agent3

Cursor

Cursor IDE / Cursor Editor / Cursor AI

VS Code-based AI editor with inline completion, chat, and Agent mode that references the entire codebase as context for refactoring.

AI Agent3

Devin (Cognition)

Devin AI / Cognition Devin / 世界初AIソフトウェアエンジニア

Cognition's autonomous AI software engineer that operates browser, terminal, and editor independently for long-horizon multi-step tasks.

AI Agent3

DocDD (Document-Driven Development)

Document-Driven Development / ドキュメント駆動開発 / DocDD手法

Design-first approach for AI agent collaboration: write structured specs before coding to prevent vibe-coding rework cycles.

AI Agent3

Function Calling

Function Calling / 関数呼び出し / ファンクションコーリング

OpenAI's term for LLM-to-function invocation — equivalent to Tool Use. Model emits a JSON call from a schema; host executes it.

AI Agent3

GitHub Copilot Workspace

Copilot Workspace / GitHub Copilot Workspace / GitHub CW

GitHub's issue-driven AI agent environment covering brainstorming, planning, code generation, testing, and automatic PR creation.

AI Agent3

MCP (Model Context Protocol)

MCP / Model Context Protocol / モデルコンテキストプロトコル

Anthropic's open protocol (Nov 2024) connecting AI agents to external tools — the LSP for AI, with a clear host/server separation.

AI Agent3

Multi-Agent

Multi-Agent / マルチエージェント / マルチエージェントシステム

Architecture where multiple AI agents collaborate — an orchestrator delegates to sub-agents, enabling work no single agent could handle.

AI Agent3

Open Interpreter

Open Interpreter CLI / 01 Light / オープンインタープリタ

Open-source agent that lets LLMs run Python, Shell, and JS in the terminal — a local Code Interpreter with autonomous file and web access.

AI Agent3

OpenAI Codex

OpenAI Codex CLI / Codex Agent / Codex Pro

OpenAI's cloud coding agent that connects to GitHub, runs tasks async in a sandbox, and auto-opens PRs. Pro tier: 100-200 USD/month.

AI Agent3

OpenAI Symphony

OpenAI Symphony / Symphony / シンフォニー

OpenAI's 2026 open-source orchestration spec — integrates with ticket tools like Linear to auto-trigger agents and submit PRs.

AI Agent3

Roo Code

Roo Code Editor / Roo Cline / RooCode

Open-source VS Code agent forked from Cline, adding custom role modes, extra tools, and enhanced MCP integration for autonomous coding.

AI Agent3

Sub-agent

Sub-agent / サブエージェント / 子エージェント

An agent spawned by an orchestrator to handle a specific subtask with its own context and git worktree.

AI Agent3

Ticket-Driven Development

Ticket-Driven Development / チケット駆動開発 / TDD（チケット駆動）

Workflow where ticket creation auto-triggers AI agents to code, test, and open a PR — humans focus on writing tickets and reviewing PRs.

AI Agent3

Tool Use

Tool Use / ツール利用 / ツール呼び出し

LLM ability to invoke external functions by emitting structured JSON — Anthropic's term for what OpenAI calls Function Calling.

AI Agent3

Vibe Coding

Vibe Coding / バイブコーディング / 雰囲気コーディング

Dev style coined by Karpathy (2025) — describe intent to an AI agent rather than write code, iterating conversationally to build fast.

AI Agent3

Voice Agent

音声エージェント / Voice AI Agent / 音声AIエージェント

AI agent using speech as its primary interface, combining real-time STT and TTS for phone support automation and voice command workflows.

AI Agent3

Windsurf

Windsurf Editor / Codeium Windsurf / ウィンドサーフ

Codeium's AI editor with a Flow agent that tracks developer actions and proactively suggests next steps for context-persistent coding.

AI Agent3

Worktree Isolation

Worktree Isolation / Worktree隔離 / gitワークツリー隔離

Using git worktree to give each parallel AI agent its own working directory — preventing conflicts in concurrent multi-agent development.

Other Categories

AI(45)AI Role(2)AI Model(19)AI Provider(6)SEO(25)Web Frontend(30)Backend(25)Database(20)Infrastructure(25)Mobile(15)Game(10)Tooling(10)