uag 0.4.21

A powerful tool execution agent

uag (Local AI Agent)

uag is a local interactive agent that executes commands, manipulates files, and reads data files such as PDF, PPTX, and Excel. It provides three user interfaces: CLI, GUI, and Web. uag is built to keep you free from vendor-locked apps: use the interface that fits your workflow, switch providers, and stay in control of your environment. GitHub: https://github.com/awaku7/agentcli

Installation

Install from PyPI with pip:

pip install uag

If you use a virtual environment, activate it first and then run the command above.

On first launch, uag checks your environment and starts the setup wizard automatically when required provider variables are missing. For configuration details, see ENVIRONMENT.md.

Key Features

• Practical toolset: File manipulation, web search, PDF/PPTX/Excel extraction, image generation, and image analysis.
• Multi-provider support: OpenAI / Azure / Bedrock / OpenRouter / Ollama / Gemini / Vertex AI / Claude / Grok / NVIDIA.
• Three interfaces:
  • CLI: uag / python -m uagent
  • GUI: uagg / python -m uagent.gui
  • Web: uagw / python -m uagent.web
  • A2A server: uaga / python -m uagent.a2a.server
• MCP support: Connect to external MCP tool servers.
• Session continuity: Keep context when switching models or providers.
• Web Inspector: Save browser transitions, DOM snapshots, and screenshots with playwright_inspector.
• Built-in docs: Read bundled docs with uag docs.

Usage

Start and exit

Run uag in your terminal to start. Type :exit to quit.

A2A server

Launch an Agent2Agent-compatible HTTP server:

uaga

See ENVIRONMENT.md for UAGENT_A2A_* settings such as auth, host, port, reload, public base URL, concurrency, and engine.

Handy tips

• :tools: show loaded tools
• :logs [n]: show recent session logs
• :load <index>: load a previous session
• :skills: select and load Agent Skills
• :shrink [n]: summarize history and keep the last n messages

Configuration and details

Environment variables and setup

For API keys, language settings (UAGENT_LANG), history shrink settings, and more, see ENVIRONMENT.md.

• Setup wizard: python -m uagent.setup_cli
• Encrypted environment: use uag_envsec to encrypt .env as .env.sec
• Update encrypted values: uag_envsec add --file .env.sec --key NAME --value VALUE

Responses API note

If you set UAGENT_RESPONSES=1, Responses API is used for supported providers: OpenAI / Azure / Bedrock / OpenRouter / Ollama. Gemini / Claude / Vertex AI use their native API paths and are not covered by Responses API. Image analysis via Responses is currently limited to OpenAI / Azure / Bedrock / OpenRouter. For other providers, uag falls back to the provider-specific or chat-completions path.

Developer docs and translations

• Developer docs: src/uagent/docs/DEVELOP.md
• Add locales: src/uagent/docs/ADD_LOCALE.md
• Other README translations: English / 日本語 / Deutsch / Español / Français / Italiano / 한국어 / Português / Русский / ไทย / 简体中文 / 繁體中文 / Polski / Tiếng Việt / Bahasa Indonesia / العربية / हिन्दी / Português / Svenska / Norsk bokmål / Suomi / Nederlands / Čeština / Українська / Swahili