parallel — Claude Code skill: parallel tasks and installation
This Claude Code skill runs independent tasks through subagents in parallel. It activates only when asked to parallelize work, compare attempts, or split a job into parts.
MIT 1,000 installs ★ 10 updated 2026-10-16
npx skills add pwguler/skills --skill parallel -a claude-codeWhat parallel does
parallel assigns separate tasks to subagents or starts several attempts at the same task. It has three modes: partitioning work, comparing full attempts, and covering parts of a job with a completion check. Parallel runs start only at the user’s request.
For partitioned work, each subagent gets a defined scope, context, constraints, and expected result. A worker that edits files uses its own git worktree; its changes return as a patch or branch for integration. Conflicts are resolved during integration, followed by the verify skill.
Race and coverage runs start only after the user agrees to the number of subagents and attempts. Partitioning requires independent tasks that do not touch the same files. Related failures are investigated together first.
What it helps with
- assign independent tasks to subagents
- compare several full attempts at one task
- split a job into parts with a completion check
- integrate changes from separate git worktrees
Example prompts for the agent
- Run these independent tasks in parallel: check the tests and trace the error in a separate module.
- Start several independent attempts to fix this error, and tell me how many first.
- Split this job into parts, define a completion check, and wait for my approval before starting.
How to install parallel
1Skills CLI (Vercel)
npx skills add pwguler/skills --skill parallel -a claude-code2Manually: put the files in place
~/.claude/skills/parallel/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/parallel/
1Skills CLI (Vercel)
npx skills add pwguler/skills --skill parallel -a codex2Manually: put the files in place
.agents/skills/parallel/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add pwguler/skills --skill parallel -a cursor2Manually: put the files in place
.cursor/skills/parallel/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install pwguler/skills skills/parallel2Manually: put the files in place
.github/skills/parallel/SKILL.md1Skills CLI (Vercel)
npx skills add pwguler/skills --skill parallel -a gemini-cli2Platform CLI
gemini skills install https://github.com/pwguler/skills --path skills/parallelManually: put the files in place
.devin/skills/parallel/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add pwguler/skills --skill 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 parallel
How do I install parallel in Claude Code?
Run npx skills add pwguler/skills --skill parallel -a claude-code. The skill file is at ~/.claude/skills/parallel/SKILL.md.
What does the parallel skill do?
It assigns independent work to subagents, compares multiple attempts at one task, or divides a job into parts with a completion check. It starts parallel work only when the user requests it.
How does the skill isolate subagent changes?
Workers that edit files use separate git worktrees. Their changes return as patches or branches, and the verify skill runs after integration.
Source text
SKILL.md · MIT
--- name: parallel description: Run independent tasks at once in subagents, race attempts at one task, or cover many parts of one job. Use it only when the user asks for parallel work. Use it when the user asks to run two or more independent tasks or failures in parallel. Use it also to race or compare several attempts at one task, or to cover many parts of one job. --- Run parallel work only when the user asks for it. Otherwise, run the tasks one at a time in this session. The independence of the tasks is the guarantee. Parallel runs are only the cheap way to get it. 1. Choose the mode. **Partition** is the default: independent tasks that share no files or state, with one subagent for each task. **Race** is several full attempts at one task, as [race.md](references/race.md) says. **Coverage** is many parts of one job with a done check, as [coverage.md](references/coverage.md) says. Related failures fit no mode. Investigate them together first. 2. For a race or a coverage run, state the cost first: how many subagents, and how many full attempts. Launch only after the user agrees. 3. For partition, give each worker a brief with its scope, context, constraints, and expected return. The scope is one file, one subsystem, or one problem. "Fix all the tests" is too wide for an agent. Paste the context into the brief: error messages, test names, and relevant paths. Agents get nothing from the session. The constraints say what must not change, such as "tests only" or "do not touch production code". The expected return is a summary of the root cause and the changes, so that you can review the integration. 4. Launch all the subagents at the same time. Run each worker that edits files in its own git worktree, so that no two workers share a checkout. A new worktree holds only committed files, so start the workers from a clean checkout. 5. Integrate the results when they return. A worker in a worktree gives its changes back as a patch or a branch, not in this checkout. Read each summary. Apply each patch with `git apply --3way`, or merge each branch. Resolve any conflict by intent. Then run the `verify` skill before you say that the batch is done. Its full gate covers the suite. Rules: - Partition does not fit in three cases. The failures are related, so a fix for one can fix the others. The cause of the problem is still unknown. Or the agents would touch the same files. - A race never averages different designs. When the attempts differ in shape, the brief was too loose. Rewrite it. A coverage run never starts without a done check. - The worktree and branch of a worker stay after the run, so that the worker can continue. When the batch passes `verify`, offer to remove them. Remove them only after the user agrees.
© pwguler · MIT · pwguler/skills