hjemmesidekongen/ai

dev-engine

Multi-agent development execution - task decomposition, specialist agents, parallel dispatch, quality gates, and 50+ framework knowledge skills.

dev-engine

The execution arm of the hjemmesidekongen/ai methodology. Where claude-core provides the infrastructure and verification, dev-engine handles the actual development work - decomposing tasks, dispatching specialist agents, and enforcing quality gates before anything ships.

Version: 0.2.0 · 53 skills · 6 agents · 2 commands

What It Does

Give dev-engine a task description, and it handles the rest: scans your project to understand the tech stack, decomposes the task into subtasks, assigns each to the right specialist agent, dispatches them in parallel with file-ownership isolation, and runs a 10-point quality gate before anything is marked done.

No agent marks its own work complete. No file is written to by two agents simultaneously. No task ships without verification.

Commands

CommandWhat It Does
/dev:runFull pipeline - decompose, assign, dispatch, verify. Supports --dry-run for preview.
/dev:scanScan repo for tech stack and architecture. Outputs .ai/project-map.yml with Mermaid C4 diagram.

Run the full pipeline:

/dev:run "Build user profile page with avatar upload and form validation"

Or preview without executing:

/dev:run "Build user profile page" --dry-run

Key Concepts

The Pipeline

Task Description


┌─────────────┐
│ project-mapper │  Scans repo → detects stack, maps modules
└──────┬──────┘


┌──────────────┐
│ taskflow-bridge │  Loads acceptance criteria (if taskflow active)
└──────┬───────┘


┌───────────────┐
│ task-decomposer │  Breaks task into ordered subtasks with dependencies
└──────┬────────┘


┌────────────────┐
│ tier-assignment  │  Maps each subtask to junior/senior/principal model
└──────┬─────────┘


┌──────────────────┐
│ context-assembler  │  Packages relevant knowledge per agent (8k token budget)
└──────┬───────────┘


┌──────────────────┐
│ agent-dispatcher   │  Parallel execution with file ownership enforcement
└──────┬───────────┘


┌────────────────┐
│ completion-gate  │  10-point quality check by code-reviewer agent
└────────────────┘

Parallel Execution with File Ownership

When dev-engine dispatches multiple agents, each one gets exclusive ownership of specific files:

frontend-dev  →  owns: src/components/Profile.tsx, src/styles/profile.css
backend-dev   →  owns: src/api/profile.ts, prisma/schema.prisma
test-engineer →  owns: tests/profile.test.ts, tests/api/profile.test.ts

No two agents write to the same file. If a task decomposition creates file conflicts, the dispatcher serializes those tasks instead of parallelizing them.

Interface contracts are defined at boundaries before implementation starts - agents agree on the API shape before building both sides.

Tier-Based Model Assignment

Not every subtask needs the most expensive model. dev-engine uses a 5-factor risk matrix to assign tiers:

FactorWeightWhat It Measures
Scope20%How many files/modules affected
Reversibility20%How hard to undo if wrong
Ambiguity25%How clear are the requirements
Impact20%Blast radius of mistakes
Dependencies15%How many other tasks depend on this

Junior (haiku): Simple, well-defined tasks - rename a variable, add a CSS class, write a straightforward test.

Senior (sonnet): Typical development work - build a component, wire up an API endpoint, implement business logic.

Principal (opus): High-risk or complex tasks - architecture decisions, security reviews, design evaluations that affect multiple modules.

10-Point Completion Gate

Every task passes through these checks before it's marked done:

  1. Spec compliance - does it match the requirements?
  2. Lint - no warnings, no errors
  3. Type check - TypeScript strict mode passes
  4. Unit tests - written and passing
  5. Integration tests - component interactions verified
  6. E2E tests - user flows tested
  7. Test coverage - meets threshold
  8. Visual verification - responsive at 3 breakpoints
  9. Security - no OWASP Top 10 violations
  10. Artifacts - all deliverables present

The code-reviewer agent runs this gate. It's the only agent authorized to approve work. Implementing agents cannot mark their own tasks done.

Component Reference

Agents

Six agents, each with a defined role and expertise:

AgentModelSpecialty
frontend-devsonnetReact, Next.js, Vue, Nuxt, CSS, accessibility, responsive design
backend-devsonnetNestJS, Prisma, Node.js, APIs, database schema, data layer
test-engineersonnetUnit, integration, E2E tests across all frameworks
architectopusDesign review, dependency analysis, module boundaries (read-only)
app-security-auditoropusOWASP checks, dependency audit, auth review, secret detection (read-only)
code-revieweropus10-point completion gate - the only agent that can approve work as done (read-only)

The three read-only agents (architect, security, code-reviewer) never modify code. They analyze and report. This separation prevents conflicts of interest - the agent that writes the code never grades it.

Framework Knowledge (53 Skills)

dev-engine doesn't give generic advice. Each framework skill contains verified, version-specific knowledge that agents load as context before working.

Frontend

SkillCoverage
react-patternsHooks, context, composition, performance, React 19 features (use(), Actions, useOptimistic)
nextjs-app-routerServer components, route handlers, caching, layouts, server actions
vue-patternsComposition API, reactivity, Pinia, provide/inject, compiler macros
nuxt-patternsAuto-imports, useFetch, server routes, layers, module development
tailwind-v4CSS-first config, @theme, @utility, container queries, dark mode
accessibility-wcagWCAG 2.1 AA, ARIA, keyboard navigation, semantic HTML, color contrast
storybook-patternsCSF3, args, play functions, interaction testing, Chromatic CI

Backend

SkillCoverage
nestjs-patternsModules, providers, guards, interceptors, pipes, exception filters
prisma-patternsSchema design, migrations, relations, N+1 elimination, transactions
nodejs-patternsStreams, worker threads, event loop, clustering, Node 22+ features

Infrastructure & Tooling

SkillCoverage
typescript-modernTS 5.4+ features, branded types, module augmentation, type-level testing
vite-patternsConfig, plugins, HMR, build optimization, manual chunks
turborepo-patternsPipeline config, caching, filtering, remote cache, Docker pruning
eslint-configFlat config (ESLint 9), custom rules, TypeScript integration
package-managerspnpm workspaces, lockfile management, peer deps, Node version pinning
openapi-typescriptCode generation, type-safe clients, contract-first development

Testing

SkillCoverage
e2e-testing-patternsPlaywright/Cypress, page objects, fixtures, network interception, visual regression
playwright-testingLocators, assertions, fixtures, visual comparison, CI sharding, trace viewer
tdd-workflowRed-green-refactor enforcement with framework-specific guidance
systematic-debugging4-phase protocol: evidence, patterns, hypotheses, validated fix

Security

SkillCoverage
security-auditOWASP Top 10, CVE scanning, secret detection, CSP headers
auth-patternsOAuth 2.0 + PKCE, JWT (RS256/ES256), refresh rotation, RBAC/ABAC
api-securityRate limiting, input validation (zod/joi), CORS, Helmet.js
web-security-owaspXSS, CSRF, SQL/NoSQL injection, SSRF, clickjacking

Expo (React Native) - 12 Skills

Full coverage for React Native development with Expo:

  • Deployment (EAS Build, OTA updates, app store submission)
  • Upgrading (SDK migration, breaking changes, rollback)
  • Native UI (platform components, gestures, animations)
  • API Routes (server functions in Expo Router)
  • Data Fetching (React Query, SWR, offline support)
  • Dev Client (custom builds, native module debugging)
  • Tailwind (NativeWind integration)
  • DOM Components (web components in native, WebView patterns)
  • SwiftUI integration (native iOS views via Expo modules)
  • Jetpack Compose (native Android views via Expo modules)
  • CI/CD (GitHub Actions, EAS, automated pipelines)
  • Simulators (iOS Simulator, Android Emulator management)

Quality & Integration

SkillCoverage
code-reviewBidirectional review with severity classification
verification-before-completionStack-specific verification commands
visual-verificationPlaywright screenshots at 3 breakpoints with LLM comparison
error-handling-patternsCustom error hierarchy, retry strategies, error boundaries
git-advanced-workflowsScripted bisect, monorepo patterns, pre-commit hooks
package-auditnpm audit, outdated dependency triage, license compliance
skill-quality-rubric8-dimension quality framework for skill auditing

Integration with taskflow

When taskflow is installed, dev-engine automatically:

  • Loads acceptance criteria from the active task
  • Maps criteria to completion gate checks
  • Uses task context to inform decomposition

When taskflow is not installed, dev-engine works standalone using only the task description you provide. No hard dependencies.

See also

  • claude-core - the foundation plugin that dev-engine depends on
  • taskflow - task management that feeds acceptance criteria into the pipeline
  • Workflows - recipes showing dev-engine in action

On this page