do-in-parallel — Claude Code skill: what it does, how to install
Runs isolated tasks across separate files concurrently via sub-agents and validates work with automated judges. Fits batch refactoring and doc generation.
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a claude-codeWhat do-in-parallel does
The skill turns Claude Code into an orchestrator for batch changes. The supervisor does not touch source files directly. In fact, editing files directly breaks the workflow. Instead, it groups requirements, assigns targets, and starts parallel implementation workers using the task tool. Model tiers (haiku, sonnet, or opus) are chosen per target, unless overridden with the --model flag.
Evaluation happens in two steps. First, meta-judges draft evaluation rubrics for each task group. Then, separate judge sub-agents score the actual output against those rubrics. If verification fails, the orchestrator triggers a retry with the judge's feedback, up to three retries per target. Setting the --strict flag rejects any result scoring below 4.0.
This architecture pays off on decoupled files. Sequential runs on identical tasks waste time; parallel workers with fresh contexts solve that bottleneck.
What it helps with
- apply identical refactoring across several files at once
- generate documentation for separate components simultaneously
- run concurrent code analysis across distinct modules
- validate sub-agent changes automatically with judge prompts
Example prompts for the agent
- /do-in-parallel Refactor error handling --files "src/auth.ts,src/user.ts" --strict
- /do-in-parallel Generate JSDoc for exported functions --targets "api,models" --model sonnet
- /do-in-parallel Check typing across components --files "components/Button.tsx,components/Input.tsx"
How to install do-in-parallel
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a claude-code2Manually: put the files in place
~/.claude/skills/do-in-parallel/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/do-in-parallel/
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a codex2Manually: put the files in place
.agents/skills/do-in-parallel/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a cursor2Manually: put the files in place
.cursor/skills/do-in-parallel/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install NeoLabHQ/context-engineering-kit do-in-parallel2Manually: put the files in place
.github/skills/do-in-parallel/SKILL.md1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a gemini-cli2Platform CLI
gemini skills install https://github.com/NeoLabHQ/context-engineering-kit --path skills/do-in-parallelManually: put the files in place
.devin/skills/do-in-parallel/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about do-in-parallel
How to install do-in-parallel in Claude Code?
Run npx skills add NeoLabHQ/context-engineering-kit --skill do-in-parallel -a claude-code.
What does do-in-parallel do?
It splits decoupled tasks among parallel sub-agents, assigns models per target, and verifies changes using LLM evaluators.
Why is the orchestrator forbidden from editing code?
The supervisor's only job is coordinating workers and judges. Reading or modifying implementation files directly breaks the parallel agent workflow.
