Agents
Specialist personas, isolated workers, or coordinated teams with distinct prompts and tools
Extension Name Custom Agents and Subagents
Vendor Terms custom agents, subagents, agent configuration, plan agent, agent permissions, task dependencies, toolAliases, allowedTools, knowledgeBase resources
Scopes
Project / Repo RootUser / Home
Interfaces IDE, CLI
Availability current (explicit)
Trust Model Agents default to read-only; write tools require explicit opt-in; allowedTools with globs controls approval-free tools; trustedAgents and agentPermissions provide layered permission control; preToolUse hooks can audit/block; IDE subagents: steering and MCP work, Specs and Hooks do not
Notes CLI custom agents are JSON config files with extensive configuration: prompt (inline or file URI), tools, allowedTools (glob patterns), toolAliases, toolsSettings, resources (file, skill, knowledgeBase with semantic search), hooks (5 trigger types), mcpServers (with OAuth), model override. IDE subagents are Markdown files with YAML frontmatter. Two built-in IDE subagents (context gathering, general purpose). Built-in Plan agent (CLI, read-only). CLI subagents support DAG-based task dependencies, per-subagent permission control, and dedicated execution monitor. ACP support enables CLI agents in JetBrains and Zed.
Extension Name Subagents
Vendor Terms subagents, custom subagents, managed subagents, built-in subagents, plugin subagents, project subagents
Scopes
User / HomeProject / Repo RootOrganization / EnterpriseCloud / Web Session
Interfaces terminal CLI, VS Code
Availability current (explicit)
Trust Model Isolated workers with their own context, prompt, tools, and optionally model; only summary returns to main conversation; permission mode inheritance from parent; plugin subagents cannot use hooks/mcpServers/permissionMode
Notes Five priority levels: managed > CLI flag > project > user > plugin. Built-in subagents: Explore, Plan, general-purpose, statusline-setup, Claude Code Guide. Supports isolation worktrees, persistent memory, skills preloading, scoped MCP servers, hooks in frontmatter, effort levels, background mode, @-mention invocation, and subagent resumption.
Extension Name Custom Agents
Vendor Terms custom agents, agent profiles, .agent.md files, Copilot cloud agent, Copilot CLI agents
Scopes
User / HomeProject / Repo RootOrganization / Enterprise
Interfaces GitHub.com, VS Code, JetBrains, Eclipse, Xcode, CLI
Availability current (explicit)
Trust Model Specialist persona with its own instructions, tool restrictions, and context; can spin up subagents; enterprise agents override org agents
Notes VS Code, GitHub.com, and CLI supported; JetBrains, Eclipse, Xcode in public preview; enterprise-level agents via .github-private repo; agent-to-agent delegation via agent tool alias
Extension Name Subagents
Vendor Terms subagents, custom agents, parallel delegated workflows, agent threads, spawn_agents_on_csv
Scopes
User / HomeProject / Repo RootCloud / Web Session
Interfaces app, CLI
Availability current (explicit)
Trust Model Subagents inherit parent sandbox policy; approval requests surface from inactive threads; runtime overrides reapplied to children; custom agents can override sandbox mode; max_depth defaults to 1
Notes Enabled by default; IDE visibility coming soon. Ships with 3 built-in agents (default, worker, explorer). Custom agents defined as TOML files at user (~/.codex/agents/) or project (.codex/agents/) scope. Experimental spawn_agents_on_csv for CSV batch processing. Max 6 concurrent threads by default; max nesting depth of 1.