Build with FreeCoder

Agents & subagents

FreeCoder isn't a single chat thread — it's an autonomous agent that can delegate. The primary agent plans and coordinates; subagents take on focused pieces of work and run in parallel, each in its own isolated git worktree so they never step on each other's changes.

Why subagents matter

  • Speed. Explore, build, and write tests at the same time instead of one after another.
  • Isolation. Each subagent gets its own git worktree, so parallel edits can't collide.
  • Focus. A subagent gets a clean, narrow task and a fresh context, which keeps quality high.

A typical fan-out

freecoder › migrate the API to v2
  plan approved → fanning out:
   · subagent A  rewrite routes/      (worktree 1)
   · subagent B  update tests/        (worktree 2)
   · subagent C  refresh the docs     (worktree 3)
  ✓ merged · tests green

Custom agents

You can define specialized agents — a reviewer, a test-writer, a refactorer — each with its own instructions and tool access, so you get a consistent expert on tap. Manage them from the command line:

freecoder agent

Custom agents pair naturally with skills (reusable know-how) and plugins (event hooks) to build a workflow that's yours.

Subagents are where FreeCoder pulls ahead of a plain chatbot — real parallelism on real repositories, included free.