What Are the Dependencies for DeepTutor? Complete Package Stack Guide

DeepTutor organizes its Python dependencies into four hierarchical layers—Core CLI, Server, TutorBot, and Development—each defined in separate requirements/ files that build upon one another through recursive -r includes starting from requirements/cli.txt.

The HKUDS/DeepTutor repository structures its runtime into modular dependency layers that support everything from command-line agent execution to multi-platform bot deployment. Understanding what are the dependencies for DeepTutor requires examining how these requirement files chain together to provide LLM integrations, RAG capabilities, API serving, and messaging platform connectors. Each layer inherits the previous through explicit file references, ensuring version consistency across CLI tools, FastAPI servers, and autonomous TutorBot instances.

Core CLI Dependencies (requirements/cli.txt)

The foundational layer defined in requirements/cli.txt powers the interactive terminal, agent tool registry, and document ingestion pipelines. This set includes strict version pins for production-critical packages like pydantic>=2.0.0 for data validation and openai>=1.30.0 for LLM interactions.

Configuration and Templating

Environment management and dynamic prompting rely on python-dotenv>=1.0.0, PyYAML>=6.0, and jinja2>=3.1.0. These packages handle secrets loading, agent configuration files, and prompt templates rendered during runtime orchestration.

LLM Provider SDKs

Multi-provider support requires openai>=1.30.0, anthropic>=0.30.0, dashscope>=1.14.0, and perplexityai>=0.1.0, alongside tiktoken>=0.5.0 for token counting. The conditional oauth-cli-kit>=0.1.1 installs only on Python 3.11+ for specific authentication flows.

Async HTTP and Resilience Patterns

The CLI performs concurrent web searches and API calls using aiohttp>=3.9.4, httpx>=0.27.0, and requests>=2.32.2. Resilience utilities include nest_asyncio>=1.5.8 for nested event loops and tenacity>=8.0.0 for retry logic within deeptutor/runtime/orchestrator.py.

Data Models and Storage

Structured output validation depends on pydantic>=2.0.0 and pydantic-settings>=2.0.0, while local SQLite caching uses aiosqlite>=0.19.0. These support the registry patterns defined in deeptutor/runtime/registry/tool_registry.py.

RAG and Document Processing

Retrieval-Augmented Generation stacks on llama-index>=0.14.12, PyMuPDF>=1.26.0, and numpy>=1.24.0,<2.0.0 for PDF parsing and vector operations. Academic paper access uses arxiv>=2.0.0, with ddgs>=9.9.1 enabling DuckDuckGo web search capabilities.

CLI Framework and Rich Output

The terminal interface in deeptutor_cli/main.py depends on typer[all]>=0.9.0 for command routing, rich>=13.0.0 for formatted console output, and prompt_toolkit>=3.0.36 for interactive user prompts.

Server Dependencies (requirements/server.txt)

The Server layer extends the CLI foundation by importing requirements/cli.txt via -r cli.txt and adding HTTP serving capabilities. This enables the FastAPI-based web interface and WebSocket communication channels.

FastAPI and ASGI Server: fastapi>=0.100.0 provides the REST API framework, served by uvicorn[standard]>=0.24.0 with CORS and logging middleware. Real-time communication requires websockets>=12.0 for socket endpoints and python-multipart>=0.0.6 for file upload handling in multipart forms.

TutorBot Dependencies (requirements/tutorbot.txt)

TutorBot extends the Server layer (-r server.txt) to support autonomous agents across messaging platforms. This layer introduces scheduling, channel-specific SDKs, and binary serialization for distributed bot instances.

Messaging Platform SDKs include python-telegram-bot[socks]>=22.6 for Telegram, slack-sdk>=3.39.0 with slackify-markdown>=0.2.0 for Slack formatting, lark-oapi>=1.5.0 for Feishu/Lark, dingtalk-stream>=0.24.0 for DingTalk, and qq-botpy>=1.2.0 for QQ channel integration. Scheduling and utilities leverage croniter>=6.0.0 for cron-based task scheduling, loguru>=0.7.3 for structured logging, and json-repair>=0.57.0 for fixing malformed LLM outputs. Networking extensions include python-socketio>=5.16.0, msgpack>=1.1.0, python-socks[asyncio]>=2.8.0, socksio>=1.0.0, and websocket-client>=1.9.0 for proxy-aware WebSocket connections. The mcp>=1.26.0 package provides Model Context Protocol support for advanced agent capabilities.

Development Dependencies (requirements/dev.txt)

The Development set inherits from requirements/server.txt to include the full runtime stack while adding quality assurance tools. Testing requires pytest>=7.0.0 and pytest-asyncio>=0.23.0 for async test coverage. Security and CI enforce pre-commit>=3.0.0 for git hooks, bandit>=1.8.0 for static security analysis, and safety<3.0.0 for vulnerability scanning of installed packages.

Dependency Hierarchy and Installation

The inheritance chain ensures consistent versioning: CLI ← Server ← TutorBot, with Development branching from Server. Install only the layer matching your deployment target.

Install the Core CLI stack for command-line agent execution:

pip install -r requirements/cli.txt
pip install -e .
deeptutor --help

Install the Server stack for API and WebSocket access:

pip install -r requirements/server.txt
pip install -e .
deeptutor serve

Install the full TutorBot stack for multi-platform bot deployment:

pip install -r requirements/tutorbot.txt
pip install -e .
deeptutor bot start

Install development tools for contributing to the repository:

pip install -r requirements/dev.txt
pytest

Key Source Files Consuming Dependencies

Several critical modules within HKUDS/DeepTutor directly import and utilize these packages:

• deeptutor_cli/main.py – Entry point that wires Typer, Rich, and Prompt Toolkit to the command-line interface, consuming the CLI dependency set.
• deeptutor/runtime/orchestrator.py – Central orchestrator relying on OpenAI, Anthropic, AsyncIO, and Tenacity for agent execution flows.
• deeptutor/runtime/registry/tool_registry.py – Registers tool implementations using Pydantic models and LlamaIndex vector stores defined in the base CLI requirements.

Summary

• Core CLI (requirements/cli.txt) provides the foundational runtime for LLM providers, RAG pipelines, and terminal interaction via Typer and Rich.
• Server (requirements/server.txt) extends CLI with FastAPI, Uvicorn, and WebSocket support for REST API deployment.
• TutorBot (requirements/tutorbot.txt) layers on top of Server to enable Telegram, Slack, DingTalk, Lark, and QQ integrations alongside scheduling utilities.
• Development (requirements/dev.txt) includes the Server stack plus pytest, pre-commit, Bandit, and Safety for testing and security auditing.
• The inheritance model uses -r includes to maintain a single source of truth for version constraints across all deployment modes.

Frequently Asked Questions

What is the minimum dependency set to run DeepTutor commands?

The minimum viable installation requires only requirements/cli.txt, which includes typer, openai, llama-index, and pydantic. This supports the core agent runtime, document processing, and interactive terminal without HTTP serving or bot capabilities.

How do I install DeepTutor with FastAPI server capabilities?

Run pip install -r requirements/server.txt followed by pip install -e .. This installs the CLI stack plus fastapi>=0.100.0, uvicorn>=0.24.0, and websockets>=12.0, enabling the deeptutor serve command documented in deeptutor_cli/main.py.

Which dependencies enable Telegram and Slack bot integrations?

The TutorBot layer in requirements/tutorbot.txt adds python-telegram-bot>=22.6, slack-sdk>=3.39.0, and slackify-markdown>=0.2.0, along with platform-specific SDKs for DingTalk (dingtalk-stream), Lark (lark-oapi), and QQ (qq-botpy). These require the Server stack as a prerequisite.

Are development dependencies required for production deployment?

No. requirements/dev.txt installs pytest, bandit, and pre-commit for contributors and CI pipelines. Production deployments should use requirements/server.txt or requirements/tutorbot.txt depending on whether you need API endpoints or messaging bot functionality.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}