super-swarm-spark — Claude Code skill: run a plan in parallel
Turns a markdown plan into a queue of subagent tasks and keeps a rolling pool busy until the code is integrated and tested.
npx skills add am-will/codex-skills --skill super-swarm-spark -a claude-codeWhat super-swarm-spark does
The skill acts as an orchestrator for subagents. It reads a plan file, pulls out task sections such as ### T1: or ### Task 1.1:, and schedules them onto subagents with the sparky role. The pool is rolling: up to 12 workers at a time, and each freed slot gets the next pending task right away. No batch-by-batch waiting.
Before anything launches, the skill assembles a context pack for the task. That means exact files and folders to touch, the precise names of files to create, neighbouring tasks hitting the same paths, and naming rules from the plan. It exists because parallel agents drift on filenames. A subagent that wants a path outside its pack has to stop and report before creating anything.
Every finished subagent gets checked, the plan file is updated, and once the queue drains the orchestrator runs an integration pass: reconcile conflicts, close test gaps, run the tests, fix what fails. Keep in mind all agents share one repository, so each is told to commit only its own files and never push. You need a plan file ready first; nothing starts without it.
What it helps with
- split a large development plan into parallel tasks
- run several subagents against one codebase at once
- stop parallel agents from inventing conflicting filenames
- merge task results into code that builds and passes tests
- run only picked tasks from a plan, like T1 T2 T4
Example prompts for the agent
- Implement the plan
plan.mdusing super-swarm-spark - Run
/super-swarm-spark ./plans/auth-plan.md T1 T2 T4 - Do only tasks T3 and T7 from this plan
How to install super-swarm-spark
1Skills CLI (Vercel)
npx skills add am-will/codex-skills --skill super-swarm-spark -a claude-code2Manually: put the files in place
~/.claude/skills/super-swarm-spark/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/super-swarm-spark/
1Skills CLI (Vercel)
npx skills add am-will/codex-skills --skill super-swarm-spark -a codex2Manually: put the files in place
.agents/skills/super-swarm-spark/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add am-will/codex-skills --skill super-swarm-spark -a cursor2Manually: put the files in place
.cursor/skills/super-swarm-spark/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install am-will/codex-skills super-swarm-spark2Manually: put the files in place
.github/skills/super-swarm-spark/SKILL.md1Skills CLI (Vercel)
npx skills add am-will/codex-skills --skill super-swarm-spark -a gemini-cli2Platform CLI
gemini skills install https://github.com/am-will/codex-skills --path skills/super-swarm-sparkManually: put the files in place
.devin/skills/super-swarm-spark/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add am-will/codex-skills --skill super-swarm-spark -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 super-swarm-spark
How do I install super-swarm-spark in Claude Code?
Run npx skills add am-will/codex-skills --skill super-swarm-spark -a claude-code. The skill lands at ~/.claude/skills/super-swarm-spark/SKILL.md.
What does super-swarm-spark do?
It parses a markdown plan into tasks and hands them to sparky subagents through a rolling pool. Results are verified, then the orchestrator resolves conflicts, adjusts tests and reruns them until they pass.
How many subagents run at the same time?
12 is the ceiling. While tasks are still pending and a slot is free, the next one is launched immediately.
