v0.1.0-beta.1 — Open Source (MIT)

AI agents for
solo developers

14 SOUL agents across 5 channels. Get answers in 30 seconds instead of 30 minutes. Built on SDLC Framework 6.3.1.

View on GitHub Usage Guide
# Initialize SDLC structure in your project $ npx endiorbot init --tier STANDARD Created: CLAUDE.md, IDENTITY.md, AGENTS.md, docs/ # Start all channels (Web + Telegram + Zalo) $ npx endiorbot serve Gateway listening on ws://127.0.0.1:18790 Telegram adapter started # Talk to AI agents $ endiorbot @pm "plan payment gateway" @pm Sprint plan created with 4 tasks...
14
AI Agents
5
Channels
39
Commands
8,124+
Tests
49
ADRs

Everything a solo developer needs

Enterprise-grade AI orchestration without the enterprise overhead

🤖

14 SOUL Agents

PM, Architect, Coder, Reviewer, Tester, DevOps, and 8 more — each with specialized prompts and SDLC-aware behavior.

📱

5 Channels

CLI, Web UI, Telegram, Zalo, Desktop (Electron). Same 39 commands work everywhere.

🔄

Multi-Model Consultation

Query Claude, GPT, Gemini, Kimi in parallel. Get consolidated expert opinions in seconds.

🚦

SDLC Quality Gates

G0 through G4 quality gates with automated evidence collection and compliance scoring.

🔗

Claude Code Bridge

Launch Claude Code sessions via tmux from any channel. Control agent runs from your phone.

⚡

Gateway Resilience

PID lockfile, provider circuit breaker, OTT 60s timeout. Built to survive real-world usage.

Architecture

Unified serve — one process, all channels, SDLC governance

User Input ──▶ Channel Adapter ──▶ MessageBus ──▶ GatewayIngress │ ┌───────────┴───────────┐ │ │ /command @agent mention │ │ CommandDispatcher ChannelRouter (39 commands) (tier-aware dispatch) │ ┌───────────┬───────┴───────┐ │ │ │ Claude Code Kimi Ollama (Primary) (Fallback) (Local LLM) Channels: CLI · Web · Telegram · Zalo · Desktop Agents: 14 SOULs × 3-tier model selection (Opus / Sonnet / Qwen) Guard: PID lockfile · Circuit breaker · Session lock

Works where you work

Same agents, same commands — any channel

💻

CLI

Terminal power users

🌐

Web UI

Browser chat at :18790

✈️

Telegram

Full OTT + bridge

💬

Zalo

Vietnam market

🖥️

Desktop

Electron native app
(Developer Preview)

14 SOUL Agents

Specialized AI personas with tier-aware model selection

AgentRoleTierUse Case
@architectSE4A ExecutorTier 1 OpusADRs, system design, G2 gate
@csoSE4H AdvisorTier 1 OpusSecurity review, threat modeling
@ceoSE4H AdvisorTier 1 OpusStrategic decisions, Go/No-Go
@coderSE4A ExecutorTier 2 SonnetCode generation, bug fixes
@reviewerSE4A ExecutorTier 2 SonnetCode review, quality checks
@pmSE4A ExecutorTier 2 SonnetSprint planning, requirements
@testerSE4A ExecutorTier 2 SonnetTest strategy, E2E coverage
@devopsSE4A ExecutorTier 2 SonnetCI/CD, deployment, infra
@fullstackSE4A ExecutorTier 2 SonnetEnd-to-end features
@researcherSE4A ExecutorTier 2 SonnetTechnology research
@ctoSE4H AdvisorTier 2 SonnetArchitecture review
@cpoSE4H AdvisorTier 2 SonnetProduct vision
@pjmSE4A ExecutorTier 2 SonnetProject tracking, velocity
@assistantRouterTier 3 OllamaGeneral queries, routing

Built on SDLC Framework 6.3.1

A proven 7-Pillar, 10-Stage AI+Human development methodology

Design Thinking Foundation
10 Lifecycle Stages
Sprint Governance
4-Tier Classification
Quality Gates (G0-G4)
SASE Integration
Documentation Permanence

sdlcframework.org — 503 methodology documents, 11 training modules (39 hours), 17 SOUL templates

View Framework on GitHub

Start building with AI agents

Free, open-source, MIT licensed. One command to get started.

⭐ Star on GitHub npm install endiorbot Read the Guide