DEV.co
AI Frameworks · HKUDS

nanobot

Nanobot is a lightweight, MIT-licensed open-source AI agent framework written in Python that lets you build personal AI assistants with web UI, multi-channel chat integration, tool calling, and memory. It supports multiple LLM providers (OpenAI, Claude, local models) and includes features like Model Context Protocol (MCP), automation, and deployment options.

Source: GitHub — github.com/HKUDS/nanobot
45.1k
GitHub stars
8k
Forks
Python
Primary language
MIT
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
RepositoryHKUDS/nanobot
OwnerHKUDS
Primary languagePython
LicenseMIT — OSI-approved
Stars45.1k
Forks8k
Open issues914
Latest releasev0.2.2 (2026-06-23)
Last updated2026-07-08
Sourcehttps://github.com/HKUDS/nanobot

What nanobot is

Python-based AI agent framework with core agent loop, WebUI (bundled), multi-channel chat adapters (Telegram, Discord, Slack, WeChat, etc.), tool execution, MCP support, model routing/fallback, memory management, and deployment via gateway. Requires Python ≥3.11 and supports OpenAI-compatible endpoints.

Quickstart

Get the nanobot source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/HKUDS/nanobot.gitcd nanobot# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Personal/Team AI Assistant Deployment

Deploy a lightweight agent across Telegram, Discord, Slack, or email with shared memory, file handling, and task automation. Useful for small teams or individual augmentation without heavy infrastructure.

LLM Multi-Provider Orchestration

Route requests across OpenAI, Claude, Anthropic, local models with automatic fallback. Practical for cost optimization, provider redundancy, or testing multiple models in production.

AI-Native Workflow Automation

Build agent-based workflows combining chat, web tools, code execution, search, and file operations. Suitable for research assistants, content pipelines, or autonomous task agents.

Implementation considerations

  • Python ≥3.11 required; use `uv pip` or standard pip for dependency installation. Bundled WebUI simplifies single-machine deployment.
  • Multi-provider setup demands API keys (OpenAI, Claude, Anthropic, etc.) with fallback model configuration for reliability.
  • Chat adapter integration is channel-specific (Telegram webhooks, Discord bots, Slack apps); each requires separate credential/permission setup.
  • MCP and tool definitions need JSON/YAML configuration; custom tools require Python function wrapping and schema definition.
  • Memory and session persistence stored locally by default; scale considerations for long-running agents or multi-user scenarios not clearly documented.

When to avoid it — and what to weigh

  • High-Security/Compliance-Sensitive Environments — Data handling, encryption, audit logging, and compliance certifications are not clearly documented. Avoid for regulated industries (finance, healthcare, PII-heavy) without security review.
  • Enterprise-Grade SLA Requirements — No mention of uptime SLAs, commercial support contracts, or incident response. Not suitable if you need guaranteed support or production-critical deployments.
  • Minimal Code/Configuration Tolerance — Setup requires Python environment, dependency management (via uv/pip), and configuration files. Non-technical users should use the documented 'Start Without Technical Background' path; others may find it complex.
  • Isolated/Air-Gapped Deployments — Architecture assumes internet access for LLM providers, web tools, MCP connections, and updates. Air-gapped or offline-first environments will face limitations.

License & commercial use

MIT License: permissive, allows commercial use, modification, distribution, and private use with attribution. No warranty or liability.

MIT permits commercial use without royalty or special permission. However, no commercial support, SLA, or indemnity is documented. For production use, review the project's governance (maintainer capacity, community stability) and consider a support agreement with maintainers or community if available.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationStrong
License clarityClear
Deployment complexityModerate
DEV.co fitStrong
Assessment confidenceHigh
Security considerations

No security audit, penetration test, or vulnerability disclosure policy mentioned. API key management relies on environment variables and config files—standard practice but not hardened. Chat adapters (Telegram webhooks, etc.) and web tool access require careful rate limiting and input validation. Session/file permissions and MCP execution sandbox not clearly detailed. Recommend: threat modeling, input sanitization, API key rotation, and network isolation for sensitive use cases before production.

Alternatives to consider

LangChain / LangGraph

Mature Python frameworks for LLM app composition; larger community, more integrations, but heavier and more complex for simple agents.

OpenAI Assistants API

Managed, hosted alternative with file/tool integration; vendor-locked to OpenAI, but no local infrastructure or deployment needed.

Anthropic's Agents (Claude + tool use)

Lightweight, focused on Claude; simpler but less flexible for multi-provider or local-model scenarios than nanobot.

Software development agency

Build on nanobot with DEV.co software developers

Nanobot is ideal for teams wanting lightweight, flexible AI automation without vendor lock-in. Start with the Quick Start guide, review security and deployment considerations for production, and evaluate community support fit for your use case.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

nanobot FAQ

Can I run nanobot fully offline?
Partially. The agent core and WebUI can run locally, but most integrations (chat platforms, web tools, LLM providers) require internet. Local LLM support via OpenAI-compatible endpoints is possible but not pre-packaged.
What LLM providers does nanobot support?
OpenAI, Anthropic (Claude), Kimi, Mistral, Moonshot, local models via OpenAI-compatible APIs. Fallback/routing to multiple providers is supported.
Is there commercial support or SLA?
Not documented. Community support via Discord, Feishu, WeChat. For production SLAs, contact maintainers directly or review commercial partnerships (Kimi, MiniMax mentioned as partners).
How do I secure API keys and sensitive data?
Keys are loaded from environment variables and config files. Use OS-level secrets management, encrypted config vaults, or container-orchestrated secret injection for production. Audit logging and data retention policies are not detailed.

Custom software development services

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If nanobot is part of your ai frameworks roadmap, our team can implement, customize, migrate, and maintain it.

Ready to Build Your AI Agent?

Nanobot is ideal for teams wanting lightweight, flexible AI automation without vendor lock-in. Start with the Quick Start guide, review security and deployment considerations for production, and evaluate community support fit for your use case.