⌨ Claude Code Ecosystem

Orchestrates large-scale changes across a codebase in parallel. Researches the codebase, decomposes the work into 5 to 30 independent units, and presents a plan. Once approved, spawns one background subagent per unit in an isolated git worktree. Each subagent implements its unit, runs tests, and opens a pull request.

Usage

/batch migrate src/ from Solid to React
/batch add JSDoc to every exported function in lib/

Requirements

• Git repository required (worktrees are how isolation is achieved)
• Review the decomposition plan before approving — each unit should be independently mergeable
• Each subagent runs /verify on its unit before opening a PR

Loads up-to-date Claude API reference material for your project's language (Python, TypeScript, Java, Go, Ruby, C#, PHP, or cURL) and Managed Agents reference. Also activates automatically when your code imports anthropic or @anthropic-ai/sdk.

Sub-commands

• /claude-api migrate — scan files for old model IDs and update to new ones; handles thinking config and parameter changes between versions
• /claude-api managed-agents-onboard — interactive walkthrough to create a new Managed Agent from scratch

Covers

• Tool use, streaming, batches, structured outputs
• Common pitfalls per language SDK
• Model IDs and their parameters

Reviews the current diff for correctness bugs and reuse/simplification/efficiency cleanups. Four parallel review agents run locally; ultra fans out to a cloud sandbox for a deeper read.

Effort levels

Flags

• --fix — apply all findings to working tree automatically
• --comment — post findings as inline GitHub PR comments
• [target] — specific path or PR reference to review

Enables debug logging for the current session and troubleshoots issues by reading the session debug log. Debug logging is off by default unless you started with claude --debug, so running /debug mid-session starts capturing logs from that point forward.

Usage

/debug                          # enable logging, run auto-diagnosis
/debug high memory usage        # focus analysis on a specific issue

Scans your transcripts for common read-only Bash and MCP tool calls, then adds a prioritized allowlist to the project .claude/settings.json to reduce permission prompts on subsequent runs. Only approves patterns Claude already performs in your sessions — nothing new is added.

When to run

• After several sessions in a new project where permission prompts interrupt flow
• When setting up a project for team members who don't want to approve common read operations

Runs a prompt repeatedly while the session stays open. Omit the interval and Claude self-paces between iterations. Omit the prompt and Claude runs an autonomous maintenance check (or the prompt in .claude/loop.md if present).

Usage

/loop 5m check if the deploy finished
/loop 10m run tests and report failures
/loop                              # autonomous maintenance mode

Launches and drives your project's app to see a change working in the running application, not just in tests or type checks. Claude builds the project, starts the server or binary, exercises the feature, and reports what it observed.

How it knows your project

Run /run-skill-generator once to teach Claude how to build and launch your specific project. The generator writes a per-project skill file that /run and /verify use on every subsequent call.

Requires

Claude Code v2.1.145 or later.

Teaches /run and /verify how to build, launch, and drive your project's app from a clean environment. Writes a per-project skill file in .claude/skills/. Run this once when you first use /run on a new project.

What it creates

• Detect build system (Go, Node, Python, etc.)
• Build command + start command
• How to verify the app is running and ready
• Project-specific interaction patterns

Reviews changed code for cleanup opportunities and applies the fixes. Four review agents run in parallel, each covering a different dimension: reuse of existing helpers, simplification of logic, efficiency improvements, and whether the change sits at the right level of abstraction.

Key distinction from /code-review

/simplify does not hunt for correctness bugs — it only improves code structure and clarity. Use /code-review for bug finding.

Requires

Claude Code v2.1.154 or later. On earlier versions, /simplify is equivalent to /code-review --fix.

Confirms a code change does what it should by building the project's app, running it, and observing the result — rather than relying on tests or type checks alone. Pairs with /run: use /run to explore, /verify to confirm a specific change works.

Difference from running tests

Tests verify code correctness. /verify verifies feature correctness — it actually drives the running application and observes behavior, catching issues tests can't see (CSS regressions, interaction bugs, environment differences).