Verno — AI-Powered SDLC Automation Engine for VS Code
Voice-first, multi-agent AI assistant that automates the full Software Development Life Cycle inside VS Code.

Architecture
graph TB
subgraph User Interface
SB[Sidebar Panel]
VP[Voice / Text Input]
SDLC[SDLC Wizard Panel]
end
subgraph Router
RT{Router}
end
subgraph Conversational Mode
CE[ConversationEngine]
WI[WorkspaceIntelligence]
end
subgraph SDLC Pipeline Mode
DO[DebateOrchestrator]
OA[OrchestratorAgent]
end
subgraph BMAD Agents
AN[Analyst - Mary]
AR[Architect - Winston]
UX[UX Designer - Sally]
PM[Product Manager]
DEV[Developer - Amelia]
CR[Code Reviewer]
QA[QA Engineer]
TW[Tech Writer]
end
subgraph AI Persona Debate
PMD[PM Persona]
ARD[Architect Persona]
FED[Frontend Persona]
QAD[QA Persona]
BED[Backend Persona]
DEVX[DevOps Persona]
SEC[Security Persona]
end
subgraph LLM Service
LS[LLMService]
GEM[Gemini Provider]
GRQ[Groq Provider]
ANT[Anthropic Provider]
OAI[OpenAI Provider]
MOCK[MockLLMProvider]
end
subgraph Cross-Cutting Services
FS[FileService]
VS[VectorStore + RAG]
FB[FeedbackService]
PS[PlanStateService]
CS[ConversationService]
TTS[TTS - Kokoro]
STT[STT - Whisper]
end
VP --> RT
SB --> RT
SDLC --> DO
RT -->|chat| CE
RT -->|plan/code| OA
CE --> WI
CE --> LS
DO --> PMD & ARD & FED & QAD & BED & DEVX & SEC
DO -->|PRD| OA
OA --> AN & AR & UX & PM & DEV & CR & QA & TW
OA --> PS
DEV --> LS
DEV --> FS
AN & AR & UX & PM & QA & TW --> LS
LS --> GEM & GRQ & ANT & OAI & MOCK
9-Phase SDLC Pipeline
Verno maps the complete Software Development Life Cycle to autonomous AI agents:
| # |
SDLC Phase |
Verno Agent / Service |
Key Output |
| 1 |
Requirements |
DebateOrchestrator → PRD Generator |
PRD.md with OWASP checklist, GDPR/HIPAA flags |
| 2 |
Planning & Estimation |
PlanningAgent + EstimationAgent |
Sprint plan, Fibonacci estimates, dependency graph |
| 3 |
Architecture & Design |
ArchitectAgent + UXDesignerAgent |
ARCHITECTURE.md, tech stack, UX wireframes |
| 4 |
Implementation |
DeveloperAgent (self-healing) |
Generated code with compile→fix→retry loop |
| 5 |
Testing & QA |
QAEngineerAgent + TestGeneratorAgent |
Test suites, coverage reports |
| 6 |
CI/CD & Deployment |
GitHub Integration Service |
Actions workflows, Dockerfiles, K8s manifests |
| 7 |
Monitoring |
OTel + Grafana + Runbook Services |
OpenTelemetry hooks, dashboard JSON, ops runbook |
| 8 |
Security |
OWASP Generator + Secret Scanner |
Compliance checklists, pre-commit hook |
| 9 |
Documentation |
ReadmeSyncService + ChangelogGenerator |
Auto-synced README, CHANGELOG.md, JSDoc |
Key Features
🎙️ Voice-First Interface
- Kokoro TTS — AI speaks responses aloud with configurable voice and speed
- Whisper STT — Hybrid local/cloud transcription with audio sanitization
- Audio Router — Auto-selects fastest transcription path
🤖 AI Persona Debates
- 7 AI personas (PM, Architect, Frontend, QA, Backend, DevOps, Security) debate your idea over 3 rounds
- Security Persona catches OWASP and compliance issues at requirements stage
- Synthesis produces a structured Product Requirements Document (PRD)
🔄 Self-Healing Code Generation
- DeveloperAgent generates code, then runs a 4-step quality pipeline:
npm install → 2. tsc --noEmit → 3. npm test → 4. npm run lint
- On failure, errors feed back to the LLM for automatic fix (up to 3 retries)
- ConflictResolverAgent merges competing agent edits via semantic merge
💬 Streaming Markdown Chat
- Real-time token streaming with
marked.js + highlight.js
- Syntax-highlighted code blocks, lists, bold, and inline code
- Provider/model selection dropdown with
localStorage persistence
🏗️ CI/CD Scaffold Generation
- Detects repo stack and generates GitHub Actions workflow
- Produces
Dockerfile, docker-compose.yml, and Kubernetes manifests
- Sprint auto-planner distributes tasks by team capacity
📊 Monitoring & Observability
- Generates OpenTelemetry instrumentation hooks per service
- Produces Grafana dashboard JSON with pre-wired panels
- Auto-generates operational runbooks from architecture context
🔒 Security & Compliance
- OWASP Top 10 checklist generated per PRD feature
- GDPR/HIPAA auto-flagging for sections with PII or health data
- Pre-commit secret scanner with git hook injection
📝 Documentation Automation
- README Auto-Sync — detects stale sections on file save, offers to regenerate
- Changelog Generator — parses Conventional Commits into
CHANGELOG.md
- JSDoc Generator — auto-documents exported functions in BMAD-generated files
🧪 Test Infrastructure
MockLLMProvider for deterministic agent testing without API calls
- Unit tests for DeveloperAgent, OrchestratorAgent, ConversationEngine
- GitHub Actions CI: lint → compile → test on every push
Installation
From VSIX (Recommended for FYP Evaluation)
code --install-extension verno-1.0.0.vsix
From Source
git clone https://github.com/maybeanns/Verno.git
cd Verno
npm ci
npm run compile
# Press F5 in VS Code to launch Extension Development Host
Configuration
All settings are under the verno.* namespace in VS Code Settings:
| Setting |
Default |
Description |
verno.defaultProvider |
gemini |
LLM provider: gemini, groq, anthropic, openai |
verno.ttsEnabled |
true |
Enable spoken responses |
verno.ttsVoice |
af_heart |
Kokoro voice model |
verno.ttsSpeed |
1.1 |
TTS playback speed multiplier |
verno.whisperMode |
hybrid |
hybrid, local-only, cloud-only |
verno.localWhisperModel |
base |
Local Whisper model: tiny, base, small, medium, large-v3 |
API keys are stored securely via VS Code SecretStorage — never in plaintext settings.
Commands
| Command |
Shortcut |
Description |
Verno: Process User Input |
Ctrl+Shift+; |
Open text input prompt |
Verno: Start Recording |
Ctrl+Shift+R |
Begin voice recording |
Verno: Start SDLC Pipeline |
Ctrl+Shift+L |
Launch debate → PRD → code pipeline |
Verno: Toggle Chat / SDLC Mode |
Ctrl+Shift+M |
Switch between modes |
Verno: Show Verno Output |
— |
View extension logs |
Verno: New Conversation |
— |
Clear conversation history |
Verno: Clear Stored API Keys |
— |
Manage provider API keys |
Verno: Generate Code |
— |
Direct code generation |
Verno: Show Diff |
— |
View generated changes |
Verno: Manage Agents |
— |
Configure agent settings |
Verno: Generate Observability Artifacts |
— |
OTel + Grafana + Runbook |
Verno: Generate OWASP Checklist |
— |
Security compliance per PRD |
Verno: Check PRD Compliance |
— |
GDPR/HIPAA flagging |
Verno: Scan for Secrets |
— |
Manual secret scan |
Verno: Install Pre-Commit Secret Scanner |
— |
Git hook injection |
Verno: Generate JSDoc |
— |
Document BMAD agent exports |
Verno: Generate Changelog |
— |
CHANGELOG.md from git history |
Project Structure
workspace/
└── .verno/
├── conversations/ # Conversation history
├── todos/ # TODO lists by agent
├── feedback/ # Agent feedback reports
├── plan-state/ # Persisted plan state + history
├── PRD.md # Product Requirements Document
├── ARCHITECTURE.md # System architecture
├── IMPLEMENTATION.md # Generated code reference
├── QA_PLAN.md # Test plans
└── CHANGELOG.md # Auto-generated changelog
Requirements
- VS Code 1.80.0 or higher
- Node.js 18 or higher
- API key for at least one LLM provider (Gemini, Groq, Anthropic, or OpenAI)
Release Notes
1.0.0 — Full SDLC Automation Engine
14-phase milestone delivering a complete AI-powered development lifecycle:
- ✅ Secure credential storage via VS Code SecretStorage
- ✅ SDLC-aware conversational engine with workspace intelligence
- ✅ 7-agent AI debate system with Security persona
- ✅ Hardened PRD generation with ambiguity detection and versioning
- ✅ Sprint auto-planner with Fibonacci estimation and dependency graphs
- ✅ Architecture + UX design generation
- ✅ Self-healing code generation with compile→fix→retry loop
- ✅ Test generation with coverage tracking
- ✅ CI/CD scaffold (GitHub Actions, Docker, K8s)
- ✅ OpenTelemetry + Grafana + Runbook generation
- ✅ OWASP compliance + GDPR/HIPAA flagging
- ✅ Pre-commit secret scanner
- ✅ README auto-sync + Changelog + JSDoc generation
- ✅ Multi-provider LLM streaming with markdown rendering
- ✅ Agent unit tests with MockLLMProvider + GitHub Actions CI
- ✅ Onboarding wizard, keyboard shortcuts, mode toggle
- ✅ Final VSIX package for distribution
0.2.0
- Automatic code quality validation in DeveloperAgent
- Feedback system across all BMAD agents
- Real-time progress tracking
- Enhanced sidebar with TODOs, Feedback, Conversations
0.0.1
- Initial release with voice recording and agent orchestration
Contributing
See CONTRIBUTING.md for development setup, testing, and PR guidelines.
License
See LICENSE file for details.
Verno — where AI agents work together to build better software, faster. 🚀
| |