Skill70 estrellas del repoactualizado 2mo ago

subagent-driven-development

Subagent-driven development enables parallel execution of independent subtasks by launching multiple Claude instances with separate specifications simultaneously. Use this pattern when a complex task contains two or more non-interdependent workstreams that each require substantial processing time and don't need to modify shared state, allowing faster completion through concurrent execution rather than sequential processing.

Ver fuente Repositorio: openclaw-superpowers

Instalar en Claude Code

Copiar

git clone --depth 1 https://github.com/ArchieIndian/openclaw-superpowers /tmp/subagent-driven-development && cp -r /tmp/subagent-driven-development/skills/core/subagent-driven-development ~/.claude/skills/subagent-driven-development

Después abre una sesión nueva de Claude Code; el skill carga automáticamente.

Definición

SKILL.md

# Subagent-Driven Development

## When to Use

- Task has 2+ independent workstreams
- Each workstream takes >10 minutes
- Workstreams do not share mutable state

## The Pattern

1. **Decompose** - break into independent subtasks
2. **Spec each subagent** - clear, self-contained instructions
3. **Launch in parallel** - use exec with background:true
4. **Monitor** - check progress, do not micromanage
5. **Integrate** - run full test suite after all complete

## Launch Pattern

```
exec command:"claude --permission-mode bypassPermissions --print '[spec A]'" background:true
exec command:"claude --permission-mode bypassPermissions --print '[spec B]'" background:true
```

Del mismo repositorio

obsidian-syncSkill

Syncs agent daily memory and MEMORY.md to an Obsidian vault so notes are human-browsable. Use nightly or on demand.

brainstormingSkill

Structured ideation before any implementation. Use when starting any non-trivial task.

create-skillSkill

Scaffolds and validates new superpowers skills. Use when creating a new skill for this repository.

executing-plansSkill

Executes plans task-by-task with verification. Use when implementing a plan.

fact-check-before-trustSkill

Triggers a secondary verification pass for any agent output containing factual claims, numbers, dates, or named entities before the output is acted on

project-onboardingSkill

Crawls a new codebase to infer stack, conventions, and key invariants, then generates a PROJECT.md context file for the agent

pull-request-feedback-loopSkill

Handles PR review feedback by fetching comments, grouping issues, fixing one group at a time, and verifying before replies.

skill-conflict-detectorSkill

Detects skill name shadowing and description-overlap conflicts that cause OpenClaw to trigger the wrong skill or silently ignore one when two skills compete for the same intent.