Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>Claude SubagentsNew to Visual Studio Code? Get it now.
Claude Subagents

Claude Subagents

Achibold Lukhwareni

|
2 installs
| (1) | Free
Visualise and orchestrate Claude subagents — plan, review, test, document and more
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Claude Subagents

Orchestrate multiple Claude AI agents directly inside VS Code — plan, code, review, test, document and more, all from a single panel.

Requirements

  • Claude Code CLI installed and authenticated
  • A Claude Pro subscription or API key

Features

Agent Canvas

Visualise your agents as a live graph. Spawn agents, chain them together into pipelines, and watch them work in real time.

Built-in Agents

  • Planner — breaks tasks into parallelisable subtasks
  • Coder — writes and edits code
  • Code Reviewer — reviews code for bugs and improvements
  • Bug Finder — finds bugs with file and line references
  • Test Writer — writes unit and integration tests
  • Docs Writer — writes JSDoc and documentation
  • Refactor — improves code structure without changing behaviour
  • Summarizer — summarises long content
  • Git Commit — writes commit messages from staged changes
  • PR Description — writes pull request descriptions
  • Negative Space — finds what's missing from your codebase

Commander

Natural language orchestration — describe what you want and Commander spawns the right agents automatically.

Pipeline Builder

Chain agents together into reusable pipelines. Add template nodes to the canvas, connect them, and run the whole pipeline with one click.

Codebase Constitution

Write a persistent project-wide system prompt (architecture decisions, coding standards, forbidden patterns) that gets injected into every agent run.

Agent DNA

Configure tone, verbosity, focus and persona per agent type. Make your coder terse, your docs writer formal, your bug finder aggressive.

Rules Management

Add global or per-agent rules (e.g. "Always write tests", "Never use any") that are injected into every agent prompt.

Knowledge Base

Save agent outputs as reusable knowledge. Relevant entries are automatically injected into future agent runs so agents build on prior work.

Memory Viewer

Browse Claude's persistent memory files directly in the panel.

Usage

Open the panel: Ctrl+Shift+U (or Cmd+Shift+U on Mac)

Right-click any file in the editor for quick access to Review, Test, Docs, Bugs, and Refactor.

License

MIT

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft