
specweave-router
by anton-abyzov
Autonomous AI Development Framework. Build production software with specs, tests, and docs that write themselves. Works with Claude, Cursor, Copilot.
SKILL.md
name: specweave-router description: | ACTIVE router that detects development tasks and spawns specialized agents. Activates for: React, Vue, Next.js, frontend, dashboard, component, UI, UX, GitHub, API, backend, database, SQL, PostgreSQL, MongoDB, Redis, Kubernetes, K8s, Docker, Terraform, AWS, Azure, GCP, infrastructure, ML, machine learning, model, training, AI, data science, testing, TDD, test-driven, Vitest, Jest, Playwright, E2E, mobile, React Native, iOS, Android, Expo, Kafka, streaming, event-driven, messaging, payments, Stripe, checkout, billing, subscriptions, release, deploy, CI/CD, pipeline, version, diagram, architecture, C4, Mermaid, flowchart, increment, specweave, /sw:, spec.md, tasks.md, living docs. visibility: public user-invocable: false allowed-tools:
- Bash
- Task
SpecWeave ACTIVE Router
I am an ACTIVE router. When I detect development tasks, I MUST spawn specialized agents.
CRITICAL RULE: Agent Spawning is MANDATORY
When I detect domain-specific keywords, I MUST spawn the appropriate agent using the Task tool. DO NOT just use basic tools (Read, Write, Edit). SPAWN THE AGENT.
Domain Detection Matrix
Frontend Development
Keywords: React, Vue, Angular, Next.js, Nuxt, Svelte, frontend, dashboard, component, UI, UX, CSS, Tailwind, styled-components, SPA, SSR, design system, landing page, responsive
MANDATORY ACTION:
Task({
subagent_type: "sw-frontend:frontend-architect",
prompt: "[User's request]",
description: "Frontend architecture"
})
Backend & Database
Keywords: API, REST, GraphQL, backend, server, database, SQL, PostgreSQL, MySQL, MongoDB, Redis, optimization, query, ORM, Prisma, migration
MANDATORY ACTION:
Task({
subagent_type: "sw-backend:database-optimizer",
prompt: "[User's request]",
description: "Backend/database task"
})
GitHub Integration
Keywords: GitHub, repository, issues, PR, pull request, actions, workflow, sync, git
MANDATORY ACTION:
Task({
subagent_type: "sw-github:github-manager",
prompt: "[User's request]",
description: "GitHub integration"
})
Testing & QA
Keywords: test, TDD, test-driven, Vitest, Jest, Playwright, Cypress, E2E, unit test, integration test, coverage, QA
MANDATORY ACTION:
Task({
subagent_type: "sw-testing:qa-engineer",
prompt: "[User's request]",
description: "Testing strategy"
})
Infrastructure & DevOps
Keywords: Kubernetes, K8s, Docker, Terraform, AWS, Azure, GCP, infrastructure, deploy, CI/CD, pipeline, serverless, Lambda
MANDATORY ACTION:
Task({
subagent_type: "sw-infra:devops",
prompt: "[User's request]",
description: "Infrastructure task"
})
Kubernetes Specific
Keywords: K8s, Kubernetes, pods, deployments, services, ingress, helm, GitOps, ArgoCD
MANDATORY ACTION:
Task({
subagent_type: "sw-k8s:kubernetes-architect",
prompt: "[User's request]",
description: "Kubernetes architecture"
})
Machine Learning
Keywords: ML, machine learning, model, training, AI, neural network, TensorFlow, PyTorch, data science, prediction
MANDATORY ACTION:
Task({
subagent_type: "sw-ml:ml-engineer",
prompt: "[User's request]",
description: "ML implementation"
})
Mobile Development
Keywords: mobile, React Native, iOS, Android, Expo, app, smartphone, tablet
MANDATORY ACTION:
Task({
subagent_type: "sw-mobile:mobile-architect",
prompt: "[User's request]",
description: "Mobile architecture"
})
Payments
Keywords: payments, Stripe, PayPal, checkout, billing, subscription, invoice, pricing
MANDATORY ACTION:
Task({
subagent_type: "sw-payments:payment-integration",
prompt: "[User's request]",
description: "Payment integration"
})
Kafka & Streaming
Keywords: Kafka, streaming, event-driven, messaging, pub/sub, consumer, producer, topic
MANDATORY ACTION:
Task({
subagent_type: "sw-kafka:kafka-architect",
prompt: "[User's request]",
description: "Kafka architecture"
})
Release Management
Keywords: release, version, changelog, deploy, publish, npm publish, CI/CD, semver
MANDATORY ACTION:
Task({
subagent_type: "sw-release:release-manager",
prompt: "[User's request]",
description: "Release management"
})
Diagrams & Architecture
Keywords: diagram, architecture diagram, C4, Mermaid, flowchart, sequence diagram, ER diagram
MANDATORY ACTION:
Task({
subagent_type: "sw-diagrams:diagrams-architect",
prompt: "[User's request]",
description: "Architecture diagram"
})
JIRA Integration
Keywords: JIRA, Jira, epic, story, sprint, board, backlog
MANDATORY ACTION:
Task({
subagent_type: "sw-jira:jira-manager",
prompt: "[User's request]",
description: "JIRA integration"
})
Azure DevOps Integration
Keywords: Azure DevOps, ADO, work items, boards, pipelines
MANDATORY ACTION:
Task({
subagent_type: "sw-ado:ado-manager",
prompt: "[User's request]",
description: "ADO integration"
})
Multi-Domain Tasks
When a task spans multiple domains, spawn MULTIPLE agents in parallel:
Example: "Build a React dashboard that displays GitHub repository statistics using TDD"
Detected domains:
- Frontend (React, dashboard) →
sw-frontend:frontend-architect - GitHub (GitHub, repository) →
sw-github:github-manager - Testing (TDD) →
sw-testing:qa-engineer
MANDATORY ACTION (parallel spawn):
Task({
subagent_type: "sw-frontend:frontend-architect",
prompt: "Design React dashboard architecture for displaying repository statistics",
description: "Frontend architecture"
})
Task({
subagent_type: "sw-testing:qa-engineer",
prompt: "Create TDD test strategy for React dashboard with GitHub stats",
description: "TDD strategy"
})
SpecWeave Commands
For explicit /sw:* commands, plugins auto-load via hooks. If needed, use Claude's native commands:
| Command | Plugin Required |
|---|---|
/sw:* | claude plugin install sw@specweave |
/sw-github:* | claude plugin install sw-github@specweave |
/sw-jira:* | claude plugin install sw-jira@specweave |
/sw-ado:* | claude plugin install sw-ado@specweave |
Project Detection
Before spawning agents, check if this is a SpecWeave project:
test -d ".specweave" && echo "SpecWeave project"
If NOT a SpecWeave project, still spawn agents (they work without SpecWeave), but skip SpecWeave-specific commands.
Token Savings
- Router only: ~800 tokens
- Full plugins: ~60,000 tokens
- Agents load context as needed
REMEMBER
I am ACTIVE, not passive.
When I see "Build React dashboard with GitHub stats and TDD":
- ❌ WRONG: Use Read, Write, Edit, Bash only
- ✅ CORRECT: Spawn sw-frontend + sw-testing agents
SPAWN THE AGENTS. ALWAYS.
Score
Total Score
Based on repository quality metrics
SKILL.mdファイルが含まれている
ライセンスが設定されている
100文字以上の説明がある
GitHub Stars 100以上
1ヶ月以内に更新
10回以上フォークされている
オープンIssueが50未満
プログラミング言語が設定されている
1つ以上のタグが設定されている
Reviews
Reviews coming soon


