EndiorBot

agent
Guvenlik Denetimi
Uyari
Health Uyari
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 7 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This tool is a local-first AI agent orchestrator designed to help solo developers manage software development lifecycle (SDLC) workflows directly from their terminal. It coordinates 14 role-based agents across multiple communication channels to assist with project planning, building, and testing.

Security Assessment
The automated code scan reviewed 12 files and found no dangerous patterns, hardcoded secrets, or requests for risky permissions. However, as an AI orchestrator, the tool inherently interacts with external Large Language Models (LLMs) to process your prompts and generate responses. Depending on your project configuration, it may transmit local code context or sensitive data to third-party AI APIs (like OpenAI or Anthropic). There are no immediate signs of malicious local execution, but the network-based data sharing required by its design warrants standard caution. Overall risk: Low to Medium.

Quality Assessment
The project is actively maintained, with its most recent code push occurring today. It uses the standard, permissive MIT license and is distributed via npm and Docker. The repository provides extensive, high-quality documentation. The primary concern is its low community visibility; with only 7 GitHub stars, it is a very early-stage or niche tool. The developer explicitly notes that it is in Beta and APIs may change, so it is not yet production-hardened.

Verdict
Use with caution: the code itself appears safe, but early-stage maturity and inherent data transmission to third-party AI APIs require you to carefully review what local context you share.
SUMMARY

Local-first AI agent orchestrator for SDLC workflows — 14 role-based agents, 5 channels, 39 unified commands. Tool-agnostic. MIT.

README.md

EndiorBot

CI
npm version
License: MIT

Beta: APIs may change between releases. Not recommended for production use yet.

AI dev team in your terminal — 14 SDLC agents, 5 channels, answers in 30 seconds

EndiorBot is a personal AI power tool for solo developers working on enterprise-scale projects.
It integrates with Claude Code as an Agent Orchestrator, enabling @agent invocations with SDLC governance across CLI, Web, Telegram, Zalo, and Desktop channels.

Identity: Solo Developer Power Tool — not a platform, not an SDLC enforcer.

Origin: EndiorBot was ported from OpenClaw (Python, internal MTS codebase) to TypeScript in Sprint 54 (ADR-005). The OpenClaw backport (Sprint 131-132) brought exec-policy, Active Memory, and SSRF protection patterns into the current architecture. OpenClaw is owned by Minh Tam Solution; code transferred to EndiorBot is released under MIT for this project.

Methodology

EndiorBot implements SDLC Framework 6.3.1 — a 7-Pillar, 10-Stage AI+Human development methodology with 11 training modules (39h).

Documentation

Application development documentation under docs/ is written in English (SDLC 6.3.1); see the Documentation language note in docs/README.md.

Prerequisites

  • Node.js >= 20
  • pnpm (via corepack: corepack enable)

Claude Code is recommended but not required. Planning, methodology, and consultation agents work without it via Kimi (free OAuth), OpenAI, or any configured LLM. Claude Code is primary for coding workflow (@coder, @architect in PATCH mode).

Install

# Via npx (no install needed)
npx endiorbot --help
npx endiorbot init
npx endiorbot serve

# Or install globally
npm install -g endiorbot

# Or via Docker
docker run -p 18790:18790 endiorbot/endiorbot serve

Quick Start

# Initialize SDLC structure
endiorbot init                              # Auto-detect tech stack
endiorbot init --tier STANDARD              # Specify tier
endiorbot compliance check                  # Verify SDLC compliance

# Unified Serve (Web + Telegram + Zalo + Desktop gateway)
endiorbot serve                             # Start all channels
endiorbot serve --no-zalo                   # Skip Zalo adapter

# Agent Orchestration (via CLI)
endiorbot @pm "plan payment gateway"        # PM agent
endiorbot @coder --patch "fix auth bug"     # Coder agent (PATCH mode)
endiorbot @consult "Redis vs PostgreSQL?"   # Multi-model consultation

Architecture

┌─────────────────────────────────────────────────────────────────────┐
│                    EndiorBot Unified Serve                           │
│                                                                      │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌───────┐ │
│  │   Web    │  │ Telegram │  │   Zalo   │  │   CLI    │  │Desktp │ │
│  │:18790/ws │  │  @bot    │  │ Bot API  │  │endiorbot │  │ App   │ │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘  └────┬─────┘  └───┬───┘ │
│       └──────────────┼───────────┬─┘             │            │     │
│                ┌─────▼─────┐     │               │            │     │
│                │ MessageBus│◄────┴───────────────┘            │     │
│                │ (debounce │◄───────────────────────────────── ┘     │
│                │  + dedup) │                                         │
│                └─────┬─────┘                                         │
│                ┌─────▼──────────────────────┐                        │
│                │     GatewayIngress         │                        │
│                │  /commands → Dispatcher    │                        │
│                │  @agents  → ChannelRouter  │                        │
│                └─────┬─────────────┬────────┘                        │
│                      │             │                                 │
│          ┌───────────▼──┐   ┌──────▼──────┐                         │
│          │ CommandDisp. │   │ChannelRouter│                         │
│          │ (39 commands)│   │ Bridge+Cloud│                         │
│          └──────────────┘   └──────┬──────┘                         │
│                                    │                                 │
│  ┌─────────────┐  ┌───────────┐  ┌▼────────────┐                   │
│  │ Claude Code │  │   Kimi    │  │   Ollama    │                   │
│  │  Bridge     │  │ (Fallback)│  │ (local LLM) │                   │
│  │ (Primary)   │  │           │  │ (Router)    │                   │
│  └─────────────┘  └───────────┘  └─────────────┘                   │
│                                                                      │
│  Per-Chat Workspace: /repos + /focus + resolveWorkspace()           │
│  SOUL Templates: 14 agents × tier-aware model selection              │
│  PID Lockfile: single-instance guard (--force to takeover)           │
└─────────────────────────────────────────────────────────────────────┘

Channels

Channel Access Purpose
Web UI http://localhost:18790 Browser chat interface
Telegram @Endior_bot Mobile OTT
Zalo Bot Endior (zapps.me) Mobile OTT (Vietnam)
CLI endiorbot @agent "task" Terminal
Desktop Electron app (auto-starts gateway) Native desktop client

OTT Commands (Telegram / Zalo)

/link                           # Bind identity (required after restart)
/repos add myapp /path/to/repo  # Register a repo
/focus myapp                    # Set active repo for this chat
/where                          # Show current focus
@pm plan the next sprint        # Talk to PM agent
@coder fix the login bug        # Talk to Coder agent
/launch claude --as coder       # Launch Claude Code in tmux
/sessions                       # List active tmux sessions
/switch <sessionId>             # Switch active session
/help                           # Full command list (39 commands)

Agent Orchestration

14 SOUL-based agents with tier-aware model selection:

Agent Category Mode Use Case
@pm executor READ Planning, requirements, sprints
@architect executor READ Design decisions, ADRs
@coder executor PATCH Code generation, bug fixes
@reviewer executor READ Code review, quality audit
@tester executor READ Test strategy, test writing
@fullstack executor PATCH Full-stack implementation
@devops executor PATCH CI/CD, infrastructure
@pjm executor READ Project management
@researcher executor READ Research, analysis
@ceo advisor READ Strategic decisions
@cto advisor READ Technical review
@cpo advisor READ Product review
@cso advisor READ Security review
@assistant router READ General queries

Multi-Agent Routing

Messages with multiple @agents or complex tasks are automatically decomposed:

@pm @cto review the authentication module
→ GoalDecomposer splits task → MultiAgentDispatcher runs in parallel

SDLC Framework

Tier Files Stages Agents
LITE 2 (CLAUDE.md, IDENTITY.md) 4 3
STANDARD 3 (+AGENTS.md) 7 6
PROFESSIONAL 4 (+USER.md) 10 10
ENTERPRISE 6 (+TOOLS.md, HEARTBEAT.md) 11 13

Per-Chat Workspace

Different chats can focus on different repos simultaneously:

# Chat A (Telegram DM): focused on EndiorBot
/focus endiorbot
@pm check sprint status

# Chat B (Zalo): focused on a separate project
/focus myproject
@coder fix the video parser

Sprint 144 Highlights (2026-04-27)

  • PID lockfile: Single-instance guard — use --force flag to take over a running instance
  • Provider circuit breaker: Auto-trip on repeated provider failures with configurable recovery window
  • OTT 60s timeout: Hard timeout on all OTT handlers (Telegram + Zalo) to prevent hung sessions
  • Desktop app: Electron-based native client with gateway auto-start on launch
  • 39 unified commands across 5 channels (CLI, Web, Telegram, Zalo, Desktop)
  • 8,142+ tests passing

Docker

# Build
docker build -t endiorbot .

# Run
docker run -p 18790:18790 \
  -e ANTHROPIC_API_KEY=your-key \
  endiorbot

# With env file
docker run -p 18790:18790 --env-file .env endiorbot

Development

pnpm install      # Install dependencies
pnpm dev          # Watch mode
pnpm build        # Build TypeScript
pnpm test         # Run tests (8,142+ passing)
pnpm lint         # Check style
pnpm lint:souls   # Validate 14 SOUL templates

Stats

Metric Value
Tests passing 8,142+
CLI commands 39 unified
Channels 5 (CLI, Web, Telegram, Zalo, Desktop)
SOUL agents 14
AI providers 6 (Anthropic, OpenAI, Gemini, Ollama, Kimi, Groq)
SDLC framework v6.3.1
Sprint 144 (2026-04-27)

Invariants

1. THIN CLIENT PATTERN: Commands call ./endiorbot.mjs core
2. STDIN JSON FOR HOOKS: Hooks receive JSON via stdin
3. ENDIORBOT SOUL = GOVERNANCE: EndiorBot decides WHAT, Claude Code executes HOW
4. DEFAULT MODEL = SONNET: Opus only for explicit architecture decisions

Documentation

Known Limitations (beta)

  • Composio tool integration not bundled (install separately if needed)
  • API surface may change between beta releases
  • No SLA or production support guarantee
  • Single-developer tested — community feedback welcome

Links

Contributing

See CONTRIBUTING.md for development setup and guidelines.

License

MIT


EndiorBot v0.1.0-beta.1 | Solo Developer Power Tool | SDLC Framework v6.3.1 | Sprint 144 (2026-04-27)

Yorumlar (0)

Sonuc bulunamadi