JSPM

AI-Powered Multi-Agent Orchestration System
Structured vibe coding for Full Stack Development.

A structured, agent-driven orchestration system that coordinates 13 specialized AI agents across the full software development lifecycle — from requirements gathering to deployment. Every decision is traceable, every artifact is validated, every session persists.

Why Chati.dev?

The Problem

AI-assisted development today suffers from three critical issues:

1. Context Loss — AI forgets decisions across sessions, leading to inconsistent implementations
2. Planning Gaps — Jumping straight to code without structured requirements leads to rework
3. System Leakage — Users accidentally "fall out" of agent systems into generic AI mode

The Solution

Chati.dev introduces Agent-Driven Development: a pipeline of 13 specialized agents where each agent owns a specific phase, produces validated artifacts, and hands off context to the next agent. An Intelligence Layer ensures context is never lost, knowledge persists across sessions, and the user never accidentally leaves the system.

PLANNING (planning)  →  Quality Gate  →  BUILD  →  Quality Gate  →  DEPLOY
  8 agents               QA-Planning     Dev        QA-Impl        DevOps

Key Innovations

Quick Start

Install

npx chati-dev init

The wizard guides you through language and project type, then auto-configures Claude Code with the required MCPs.

Activate

/chati

The orchestrator loads your session, detects where you left off, and routes you to the right agent. You stay inside the system until you explicitly exit.

Monitor

npx chati-dev status          # One-time snapshot
npx chati-dev status --watch  # Auto-refresh every 5s

Exit & Resume

/chati exit     # Save session and exit
/chati          # Resume exactly where you left off

Architecture

13 Agents, 4 Categories

Pipeline Flow

User Request
    │
    ▼
┌─────────────────────────────────────────────────┐
│  ORCHESTRATOR (/chati) — Main Terminal           │
│  Routes to correct agent, manages session        │
│  Session Lock: user stays in system              │
└─────────────────────────────────────────────────┘
    │
    ├── In-conversation ──┐
    │                     ▼
    │   ┌─── PLANNING (interactive) ───┐
    │   │  WU → Brief                 │
    │   └─────────────────────────────┘
    │
    ├── Spawns terminals ──────────────────────────┐
    │                                               ▼
    │   ┌─── PLANNING (autonomous) ─── Parallel ───┐
    │   │  Detail (opus) ──┐                       │
    │   │  Architect (opus)├── run simultaneously  │
    │   │  UX (sonnet) ────┘                       │
    │   │  → Phases → Tasks                        │
    │   └──────────────────────────────────────────┘
    │
    ├── Spawns terminal ──┐
    │                     ▼
    │   ┌─── QUALITY ─────────────────────────────┐
    │   │  QA-Planning (opus) — ≥95% threshold     │
    │   └──────────────────────────────────────────┘
    │
    ├── Spawns terminal(s) ┐
    │                      ▼
    │   ┌─── BUILD ───────────────────────────────┐
    │   │  Dev (opus) — parallel for independent   │
    │   └──────────────────────────────────────────┘
    │
    ├── Spawns terminal ──┐
    │                     ▼
    │   ┌─── QUALITY ─────────────────────────────┐
    │   │  QA-Implementation (opus) — tests + SAST │
    │   └──────────────────────────────────────────┘
    │
    └── Spawns terminal ──┐
                          ▼
        ┌─── DEPLOY ──────────────────────────────┐
        │  DevOps (sonnet) — git + deploy + docs   │
        └──────────────────────────────────────────┘

Each spawned terminal runs as a separate claude -p --model <model> process with its own context window, write-scope isolation, and structured handoff output.

Intelligence Layer

The Intelligence Layer operates transparently behind the pipeline:

Universal Protocols

Every agent follows 8 universal protocols:

Constitution

The system is governed by a 17-article Constitution that enforces agent behavior, quality standards, security, and system integrity. Key articles:

Supported IDEs

CLI Commands

Orchestrator Commands (inside active session)

Project Structure

your-project/
├── .chati/
│   ├── session.yaml              # Session state (auto-managed, gitignored)
│   └── memories/                 # Memory Layer storage (gitignored)
├── .claude/
│   ├── commands/
│   │   └── chati.md              # Thin router → orchestrator
│   └── rules/
│       └── chati/                # Framework context (auto-loaded by Claude Code)
│           ├── root.md           # System overview + key references
│           ├── governance.md     # Constitution key rules
│           ├── protocols.md      # 8 universal protocols
│           └── quality.md        # Quality standards
├── CLAUDE.md                     # Minimal project context (auto-generated)
├── CLAUDE.local.md               # Runtime state — session lock, current agent (gitignored)
├── docs/
│   └── CHANGELOG.md              # Version history
├── chati.dev/
│   ├── orchestrator/             # Main orchestrator
│   ├── agents/                   # 13 agent definitions
│   │   ├── planning/              # 8 planning agents
│   │   ├── quality/              # 2 quality gate agents
│   │   ├── build/                # Dev agent
│   │   └── deploy/               # DevOps agent
│   ├── tasks/                    # 72 task definitions (YAML frontmatter)
│   ├── workflows/                # 5 workflow blueprints
│   ├── templates/                # 5 artifact templates
│   ├── schemas/                  # 5 JSON schemas for validation
│   ├── intelligence/             # Context Engine, Memory Layer, Decision Engine
│   ├── domains/                  # Domain loading configs (per-agent, per-workflow)
│   ├── hooks/                    # 5 Claude Code hooks (enforcement)
│   ├── context/                  # Context source files (deployed to .claude/rules/)
│   ├── frameworks/               # Decision heuristics, quality dims
│   ├── quality-gates/            # Planning & implementation gates
│   ├── patterns/                 # Elicitation patterns
│   ├── data/                     # Entity registry (48 artifacts)
│   ├── i18n/                     # EN, PT, ES, FR translations
│   ├── migrations/               # Version migration scripts
│   ├── constitution.md           # 17 Articles + Preamble
│   └── config.yaml               # System configuration
├── chati.dev/artifacts/          # Generated during pipeline
│   ├── 0-WU/
│   ├── 1-Brief/
│   ├── 2-PRD/
│   ├── 3-Architecture/
│   ├── 4-UX/
│   ├── 5-Phases/
│   ├── 6-Tasks/
│   ├── 7-QA-Planning/
│   ├── 8-Validation/
│   └── handoffs/
└── packages/
    └── chati-dev/                # CLI + runtime engine (npx chati-dev)
        └── src/terminal/         # Multi-terminal spawner, monitor, collector

Internationalization

The installer and agent interactions support 4 languages:

Artifacts are always generated in English for portability and team collaboration.

Upgrade System

npx chati-dev check-update                # Check for updates
npx chati-dev upgrade                      # Upgrade to latest
npx chati-dev upgrade --version 1.2.1      # Specific version

Upgrades include automatic backup, migrations, validation, and config merging. Rollback on failure.

Prerequisites

• Node.js >= 20.0.0
• npm >= 9.0.0
• A supported IDE with AI assistant capabilities

Contributing

We welcome contributions from agents, templates, workflows, intelligence data, translations, and CLI improvements. Please see CONTRIBUTING.md for guidelines.

Security

For security concerns, please see our Security Policy.

License

This project is licensed under the MIT License — see LICENSE for details.

_{Built with structure, validated by agents, governed by constitution.
Chati.dev © 2026}