code-simplifier — Anthropic Official Cleanup Subagent

How code-simplifier Works

Boris Cherny, Threads (@boris_cherny, post DTBVroqEg_K):

"I use a few subagents regularly: code-simplifier simplifies the code after Claude is done working, verify-app has detailed instructions for testing Claude Code end to end, and so on. Similar to slash commands, I think of subagents as automating the most common workflows that I do for most PRs."

What it actually does

Typical session flow

You:    "Implement the multi-tenant audit log feature."
Claude: ... (writes 4 files, ~600 lines)
You:    "Now run the code-simplifier on what you just changed."
Claude: -> invokes code-simplifier subagent
        -> reviews 4 files
        -> commits cleanup as a separate diff

When to use

• End of any session where Claude wrote >100 lines
• Before opening a PR — separates "feature commit" from "cleanup commit"
• After /loop or /ralph-loop runs (long sessions accumulate cruft)
• Avoid mid-feature — let Claude finish the implementation first
• Avoid on code you didn't write in this session (it's optimized for recently modified files)

Why Boris ships with it on every PR

From his Threads post: he treats subagents the same way he treats slash commands — "automating the most common workflows that I do for most PRs." code-simplifier is one of the two he names by name (the other is verify-app).

Example session

User: Build a function that takes a list of users, filters out inactive ones,
      sorts by created_at desc, and returns the top 10.

Claude: [writes a 25-line implementation with nested filters and a temp variable]

User: Run code-simplifier on it.

code-simplifier: Refactored to a single chained pipeline.
                  - Removed temp variable filtered_users
                  - Replaced lambda x: x.active == True with attrgetter('active')
                  - Inlined the sort key
                  Behavior preserved: same input -> same output, verified
                  against the existing tests.

FAQ

Q: What is code-simplifier? A: An Anthropic-published Claude Code subagent that cleans up code Claude just wrote — refactoring for clarity and consistency while preserving exact behavior. Open-sourced from the internal Claude Code team's own setup.

Q: Is code-simplifier free? A: Yes, free and open source via Anthropic's claude-plugins-official marketplace. You only pay for Claude API tokens used during the cleanup pass (it runs on Opus, so factor that in).

Q: How do I install code-simplifier? A: Run claude plugin install code-simplifier in your terminal, or /plugin install code-simplifier from inside a Claude Code session after /plugin marketplace update claude-plugins-official.

Q: Will it change my code's behavior? A: No. The agent's contract is to never change observable behavior — only how the code is structured. Public APIs, return values, and side effects all stay the same.

Q: When should I run it? A: At the end of a long coding session, before opening a PR. Don't run it mid-feature — wait until the implementation is functionally complete, then have it cleanup pass over what was changed.