AI Coding Tools Terminology Comparison

A comparison of terminology used across major AI coding tools (as of February 2026).

Comparison Table

Concept Claude Code OpenAI Codex Kiro IDE Kiro CLI Gemini CLI OpenCode
The AI itself Agent / Subagent Agent Agent + Autonomous Agent (async) Agent Agent (ReAct loop) Agent
Specialized agents Subagents (Explore, Plan, Bash, etc.) + Custom Agents Subagents + Powers (context bundles) Subagents + Custom Agents Sub-agents (experimental: cli_help, routing, generalist) + Jules (async) + Custom Agents Primary Agents (Build, Plan) + Subagents (General, Explore) + Custom Agents
Callable capabilities Tools (Read, Edit, Bash, Grep, Glob, Write, WebFetch, WebSearch, etc.) Tools (shell, file read/write, apply patch, etc.) Tools (built-in + MCP tools) Built-in Tools (read, write, shell, aws, grep, glob, subagent, web search/fetch) Tools (ReadFile, WriteFile, Edit, Shell, FindFiles, SearchText, GoogleSearch, WebFetch, SaveMemory, etc.) Tools (bash, read, write, edit, grep, glob, list, patch, lsp, webfetch, websearch, etc.)
External integrations MCP Servers MCP Servers MCP Servers (local + remote) MCP Servers (local + remote, with MCP Registry governance) MCP Servers (stdio + SSE transports) MCP Servers (local + remote with OAuth)
Reusable workflows Skills (slash commands) + Plugins Skills + /review + /plan slash commands Specs (SDD) + Agent Skills Skills (progressively loaded, on-demand) Extensions + Agent Skills (SKILL.md) + Custom Slash Commands Skills (SKILL.md files in .opencode/skills/)
Event-driven automation Hooks (14 events: PreToolUse, PostToolUse, SessionStart/End, UserPromptSubmit, Stop, SubagentStart/Stop, PreCompact, etc.) Agent Hooks (file save/create/delete, prompt submit, agent turn, pre/post tool use) Hooks (agentSpawn, userPromptSubmit, preToolUse, postToolUse, stop) Hooks (SessionStart/End, BeforeModel/AfterModel, BeforeTool/AfterTool, BeforeAgent/AfterAgent, BeforeToolSelection, PreCompress) Plugins (tool.execute.before/after, file.edited, session events, etc.)
Project-level AI instructions CLAUDE.md (hierarchical: user/project/local) AGENTS.md (hierarchical, with configurable fallbacks) Steering Files (.kiro/steering/) Steering Files (.kiro/steering/) + Agent config JSON GEMINI.md (hierarchical, multi-file, configurable filename) AGENTS.md (also reads CLAUDE.md for compatibility)
Persistent memory Auto memory (~/.claude/projects/<project>/memory/) + Subagent memory Thread memory (experimental) Autonomous Agent learns from PR feedback Knowledge management (experimental, /knowledge) save_memory (built-in tool) + Conductor (extension) — (available via community plugins)
Bundled packages Plugins (commands, agents, skills, hooks, MCP, LSP) Powers (MCP + steering + hooks) + Agent Skills + Open VSX plugins Custom Agents (bundled config + tools + steering) Extensions (MCP + context + slash commands + skills) Plugins (JS/TS modules with hooks + custom tools)
Background/async work Background tasks Automations (scheduled runs in background worktrees) Autonomous Agent (runs for days, up to 10 concurrent tasks) Subagent parallelism Jules (async in VM, up to 15 concurrent tasks) + Sub-agents — (available via community plugins)
Parallel execution Subagent parallelism Cloud worktrees Subagents (parallel) + Autonomous Agent (10 concurrent) Subagents (isolated context, live progress) Jules (15 concurrent tasks) + Sub-agents (experimental) General subagent (limited parallelism)
Planning mode Plan agent /plan slash command Spec-Driven Development + Plan Agent Plan agent (built-in, Shift+Tab or /plan) Plan Mode (experimental, read-only) Plan agent (read-only, Tab to switch)
Configuration settings.json (JSON, hierarchical scopes) config.toml (TOML, with profiles) .kiro/ directory (steering md, hooks, agents md, skills) Agent config JSON + Steering Markdown settings.json (JSON) + GEMINI.md opencode.json (JSON/JSONC)

Key Takeaways

Sources

Claude Code

OpenAI Codex

Kiro IDE

Kiro CLI

Gemini CLI

OpenCode