FlexiAI Toolsmith

FlexiAI Toolsmith is a flexible, multi-channel Python framework for building AI-powered chat assistants. It supports both CLI and web (Quart + SSE) interfaces, integrates with multiple AI providers via the OpenAI Python SDK, and enables assistants to invoke powerful tool plug-ins—including CSV / spreadsheet processing, YouTube search, security audits, dynamic forms, and (soon) OCR.

Provider Support Overview

• ✅ OpenAI and Azure OpenAI — full Assistant-API support
• 🟡 DeepSeek and Qwen — OpenAI SDK, but chat-completions only for now
• ❌ GitHub Azure Inference — chat-completions only

⚙️ When any SDK-backed provider adds Assistant-API support, FlexiAI Toolsmith will pick it up automatically—no code changes required.

---

Table of Contents

1. Features
2. Tutorials / Quick Demos
3. Architecture
4. Prerequisites
5. Installation
6. Configuration
7. Usage
   • CLI Chat (WSL or Linux)
   • Web Chat (Quart -- SSE)
8. Contributing
9. License

---

Features

Multi-Channel Publishing

Chat events can be streamed to:

• CLI (CLIChannel)
• Redis Pub/Sub (RedisChannel)
• SSE Web clients via Quart (QuartChannel + SSEManager)

---

Tutorials / Quick Demos

  

---

AI Provider Support

• OpenAI & Azure OpenAI — full Assistant-API (threads, deltas, tool calls)
• DeepSeek & Qwen — chat-completions via OpenAI SDK (Assistant-API pending)
• GitHub Azure Inference — chat-completions only

Streaming Assistant API

• Thread life-cycle management (create → queue → in-progress → complete)
• Event routing via in-memory EventBus + EventDispatcher
• Delta-based message streaming with MessageDeltaEvent

Dynamic RAG & Multi-Agent Tool Orchestration (via Toolsmith)

Toolsmith lets assistants invoke dynamic tools via tool calls, giving you a hybrid RAG + MAS system:

• Agent coordination & delegation
  • save_processed_content / load_processed_content
  • initialize_agent / communicate_with_assistant

Built-in Tool Plug-ins

• YouTube Search — search_youtube, search_on_youtube
• Product Filtering — filter_products (ai_custom_products)
• CSV Operations — csv_operations (CRUD)
• Subscriber Management — identify_subscriber, retrieve_billing_details, manage_services
• Spreadsheet Operations — file_operations, sheet_operations, data_entry_operations, …
• Security Audits — security_audit (recon, port scans, defenses, updates)
• Web Forms (Experimental) — Generates interactive forms as Markdown in the Quart/SSE chat UI; submissions are sent via a POST to /submit_user_info and persisted to CSV
• OCR (Coming Soon) — flexiai/toolsmith/_recycle/OCR.py

Configurable & Extensible

• Pydantic-based .env for credentials, channels, logging, …
• Modular factories for channels, credentials, handlers, tools
• Structured logging with rotating file + console outputs

---

Architecture

📦 flexiai-toolsmith
 ┣ 📂 flexiai
 ┃ ┣ 📂 channels
 ┃ ┣ 📂 config
 ┃ ┣ 📂 controllers
 ┃ ┣ 📂 core
 ┃ ┃ ┣ 📂 events
 ┃ ┃ ┗ 📂 handlers
 ┃ ┣ 📂 credentials
 ┃ ┣ 📂 database
 ┃ ┣ 📂 toolsmith
 ┃ ┗ 📂 utils
 ┣ 📂 static
 ┣ 📂 templates
 ┣ 📜 .env
 ┣ 📜 .env.template
 ┣ 📜 .gitignore
 ┣ 📜 app.py
 ┣ 📜 chat.py
 ┣ 📜 environment.yml
 ┣ 📜 requirements.in
 ┗ 📜 requirements.txt

---

Prerequisites

• Python 3.12+
• Redis (if redis in ACTIVE_CHANNELS)
• Conda (Miniconda/Anaconda) or pip + venv

---

Installation

1. Clone the repository

   git clone https://github.com/SavinRazvan/flexiai-toolsmith.git

2. Set up the environment (Conda recommended)

   conda env create -f environment.yml
   conda activate .conda_flexiai

   Or use venv + pip:

   python -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

---

Configuration

Copy .env.template ➜ .env, then edit:

Variable	Description	Example
CREDENTIAL_TYPE	openai, azure, deepseek, qwen, github_models	openai
OPENAI_API_KEY, …	Provider credentials	sk-…
ACTIVE_CHANNELS	cli, redis, quart (comma-separated)	cli,quart
USER_PROJECT_ROOT_DIR	Absolute project root path	/home/user/code
YOUTUBE_API_KEY (opt.)	Needed for YouTube search tool	AIza…

ℹ️ Assistant-API features currently work on OpenAI & Azure OpenAI only; DeepSeek & Qwen auto-enable once their endpoints support it.

---

Usage

CLI Chat (WSL or Linux)

python chat.py

• Prompts show as 👤 You
• Assistant messages stream as 🌺 Assistant

Web Chat (Quart + SSE)

hypercorn app:app --bind 127.0.0.1:8000 --workers 1

1. Browse to http://127.0.0.1:8000/chat/
2. Start chatting in the live SSE UI

---

Contributing

1. Fork ➜ create a feature branch

   git checkout -b feature/my-feature

2. Commit with clear messages

3. Open a pull request explaining context & purpose

---

License

Released under the MIT License — see LICENSE for full terms.