Metadata-Version: 2.4
Name: agentic-workstation
Version: 1.0.0
Summary: Multi-agent planning and workflow orchestration system
Author-email: Sujith <sujith.eag@gmail.com>
License-Expression: MIT
Requires-Python: >=3.10
Requires-Dist: click>=8.1.0
Requires-Dist: jinja2>=3.1.0
Requires-Dist: jsonschema>=4.17.0
Requires-Dist: platformdirs>=3.0.0
Requires-Dist: pyyaml>=6.0.0
Requires-Dist: rich>=13.0.0
Requires-Dist: structlog>=25.0.0
Requires-Dist: toml>=0.10.0
Requires-Dist: tomli-w>=1.0.0
Requires-Dist: tomli>=1.2.0
Provides-Extra: auth
Requires-Dist: pyjwt>=2.8.0; extra == 'auth'
Provides-Extra: dev
Requires-Dist: mypy>=1.0.0; extra == 'dev'
Requires-Dist: pytest-cov>=4.0.0; extra == 'dev'
Requires-Dist: pytest>=7.0.0; extra == 'dev'
Requires-Dist: ruff>=0.1.0; extra == 'dev'
Provides-Extra: external
Requires-Dist: requests>=2.31.0; extra == 'external'
Provides-Extra: interactive
Requires-Dist: click-completion>=0.5.0; extra == 'interactive'
Requires-Dist: questionary>=2.0.0; extra == 'interactive'
Provides-Extra: performance
Requires-Dist: cachetools>=5.3.0; extra == 'performance'
Provides-Extra: web
Requires-Dist: fastapi>=0.100.0; extra == 'web'
Requires-Dist: uvicorn>=0.23.0; extra == 'web'
Description-Content-Type: text/markdown

# Agentic Workflow System

> **Structured Architectural Scaffolding for AI Development**

[![Version](https://img.shields.io/badge/version-1.0.0-blue.svg)](https://github.com/agentic-workflow-os)
[![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
[![Docker](https://img.shields.io/badge/docker-ready-blue.svg)](pkgs/container/agentic)

**Agentic Workflow** is a development platform that orchestrates Multi-Agent Systems to plan, architect, and implement complex software projects. Unlike "Chat with Code" tools that rely on messy, unstructured conversation history, this system enforces a **Context-First** philosophy. It treats Agent Context as a file-system state machine, ensuring that your AI Engineer knows exactly what your AI Architect decided.

---

## ⚡ Why Use This?

Most AI coding tools suffer from **Context Drift**. After 20 messages, the AI forgets the architectural constraints you set in message #1.

AI for projects and work is like a rocket: it gives a quick boost of velocity if you have direction, but soon everyone realizes it's only a first-stage booster that stops midway once context is off. The velocity drops, and to carry forward and build production-grade applications needs a 2nd and 3rd stage booster with precision guidance to reach and stay in orbit. **This project is that 2nd and 3rd stage.**

**Agentic Workflow solves this by:**
1.  **Strict Role Separation:** "Architects" write specs. "Engineers" write code. They never swap roles.
2.  **Artifact-Driven Handoffs:** An agent cannot proceed until it receives a specific, validated artifact (e.g., `api_spec.md`) from its predecessor.
3.  **No "Chatting":** You don't chat in the terminal. The system generates a comprehensive **Context File** that you drop into your preferred AI (ChatGPT, Claude, Gemini, VS Code Copilot). The AI does the work, and you save the result. The system tracks the state.

---

## 🚀 Quick Start

### Installation

**Option A: Pip (Recommended)**
```bash
pip install agentic-workflow
```
*Note: Package will be available on PyPI soon. For now, install from source.*

**Option B: Docker**
```bash
# Build locally
docker build -t agentic-workflow .
alias agentic="docker run -it -v $(pwd):/data agentic-workflow"
```

### 1. Initialize a Project
Navigate to your empty folder and initialize a project.
```bash
mkdir my-new-saas && cd my-new-saas
agentic project init MySaaS
```
*This creates a project with the default "planning" workflow. Use `--workflow` to specify another workflow.*

### 2. The "Active Session" Loop
The system creates an `active_session.md` file. This is your interface with the AI.
1.  **Activate:** `agentic workflow activate A-01` (Project Guide).
2.  **Generate:** Copy the content of `active_session.md` into your LLM (Claude/GPT-4).
3.  **Save:** Paste the LLM's output into the requested file (e.g., `artifacts/A-01/project_brief.md`).
4.  **Handoff:** Tell the system you are done.
    ```bash
    agentic workflow handoff --to A-02 --artifacts artifacts/A-01/project_brief.md
    ```
    *The system validates that the file exists before allowing the handoff.*

### 3. Review Status
View the decision tree and current state.
```bash
agentic workflow status
```

---

## 📦 Workflows

The system supports pluggable **Workflow Manifests**.

### 1. Planning (Canonical)
*   **Goal:** Turn a 1-sentence idea into a full Tech Spec.
*   **Agents:**
    *   `A-01 Incubation`: Refines the idea.
    *   `A-02 Requirements`: Lists functional/non-functional reqs.
    *   `A-03 Architect`: Designs the system topology.
    *   `A-04 Security`: Threat modeling.
    *   (And 10 more specialized roles).

### 2. Implementation (Canonical)
*   **Goal:** Turn specs into code.
*   **Agents:**
    *   `E-01 Frontend`: React/Vue/Svelte implementation.
    *   `E-02 Backend`: API & Logic.
    *   `E-03 Database`: Schema & Migrations.

### 3. Custom Workflows
You can create your own workflows by placing a manifest in:
- Project: `./.agentic/workflows/my-custom-flow/workflow.json`
- User global: `~/.config/agentic/workflows/my-custom-flow/workflow.json`
- System: Bundled in package (not recommended for custom)

Workflow manifests include `workflow.json`, `agents.json`, `artifacts.json`, etc.

---

## ⚙️ Configuration & Governance

### `agentic.toml`
Every project handles its own configuration.
```toml
[project]
name = "MySaaS"
version = "0.1.0"

[governance]
require_reviews = true
git_commit_on_handoff = true
```

### Global Config (`~/.config/agentic/config.toml`)
Set your preferences for all projects.
```toml
[directories]
projects = "projects"

[governance]
level = "moderate"

[workflows]
default = "planning"
```

---

## 🛠️ Development

This project is built with Python 3.10+.

*   **Core:** `src/agentic_workflow`
*   **Manifests:** `src/agentic_workflow/manifests` (Canonical Definitions)
*   **Templates:** `src/agentic_workflow/templates` (Jinja2 templates)
*   **Build:** `pyproject.toml` (Hatchling)
*   **Tests:** Not included in deployment

**License:** MIT
