LabsAI-Driven Software Factory
A multi-agent SDLC orchestrator that coordinates Analysts, Project Managers, Developers, and QA agents across multiple LLM providers to iteratively build and refine complete applications. Five feature-parity implementations plus a Next.js web frontend with real-time SSE dashboard, run history with replay, and OIDC auth via Rdn.Identity.
PHASE 1: ANALYST
Analysis complete | Components: 4
PHASE 2: PROJECT MANAGER
Work plan created | Developers: 4
PHASE 3: DEVELOPERS
Dev 0: index.html
Dev 1: style.css
Dev 2: calculator.js
Dev 3: script.js
PHASE 4: INTEGRATION ARCHITECT
2 integration issues fixed
PHASE 5: QA REVIEW
13 issues | Critical: 4 | Major: 2
Completion Score: 64.8%
Cost: $0.14 (anthropic | openai | deepseek | google)
7 Phases
Full SDLC
4 Providers
Claude + GPT + Gemini + DeepSeek
7-Phase SDLC Pipeline
A complete software development lifecycle with specialized agents for analysis, planning, development, integration, QA, and documentation - with iterative refinement until quality thresholds are met.
Multi-Provider Support
Mix Anthropic Claude, OpenAI GPT, Google Gemini, and DeepSeek models per role. Extensible architecture makes adding new providers straightforward.
Quality-Gated Iterations
Automatic re-iteration until completion threshold is met. Configurable scoring weights for critical, major, and minor issues.
Per-Provider Cost Tracking
Real-time token usage and cost breakdown by provider and role. Set max cost limits to control spending across iterations.
5 Implementations, One Pipeline
Node.js
v24+ with ES Modules
- Promise.all/race for concurrency
- @anthropic-ai/sdk integration
- Native async/await patterns
Python
3.12+ with asyncio
- asyncio.Semaphore for limits
- Anthropic async client
- asyncio.gather for parallelism
Go
1.25+ with chromedp
- Goroutines + WaitGroup
- chromedp for browser automation
- Channel-based semaphore
.NET
.NET 10 with C#
- Task.WhenAll for parallelism
- SemaphoreSlim for limits
- Native HttpClient API
Next.js
Next.js 16 + React 19
- SSE real-time dashboard
- OIDC auth via Rdn.Identity
- Docker-ready deployment
Per-Provider Cost Breakdown
Track token usage and costs across all providers. See exactly where your budget is going with detailed breakdowns by role and provider.
💰 COST BY PROVIDER:
--------------------------------------------------
Provider Calls Tokens Cost
--------------------------------------------------
anthropic 12 44,405 $ 0.0678 (54%)
openai 6 28,100 $ 0.0342 (27%)
deepseek 4 18,200 $ 0.0126 (10%)
google 4 15,670 $ 0.0099 (8%)
--------------------------------------------------
TOTAL 26 106,375 $ 0.1245
ANTHROPIC BREAKDOWN:
API calls: 12
Input tokens: 37,315
Output tokens: 7,090
Completion Score Calculation
Weighted scoring algorithm: critical issues (50%), major (20%), minor (10%), and acceptance criteria (20%). Score capping prevents artificially high scores when critical issues remain.
Cost Limiting
Set maximum cost thresholds with --max-cost. Orchestrator stops automatically when budget is reached.
User Check-In Points
Interactive prompts between iterations showing score, issues, and cost. Accept results early or continue iterating.
7 SDLC Phases
Each phase uses specialized agents with role-specific prompts and configurable LLM providers. The pipeline iterates until quality thresholds are met.
Analyst
Requirements decomposition
PM
Work assignment
Developers
Parallel code gen
Architect
Static analysis
QA Review
Issue detection
Feedback
Score calculation
Summary
Documentation
Explore the Architecture
Dive into the technical details of multi-agent orchestration, concurrency patterns, and production-ready coordination mechanisms.