#6969th of 153 in Agent frameworks and SDKs

langroid

Python framework for multi-agent LLM apps with tools and RAG

Stars
4.2k
License
MIT
Last commit
Nov 2026
Last release
Nov 2026
Language
Python
Stars, last 29 weeks+192
25 Apr 2026: 3,961 stars2 May 2026: 3,972 stars9 May 2026: 3,980 stars16 May 2026: 3,989 stars23 May 2026: 3,999 stars30 May 2026: 4,005 stars6 Jun 2026: 4,008 stars13 Jun 2026: 4,018 stars20 Jun 2026: 4,022 stars27 Jun 2026: 4,029 stars4 Jul 2026: 4,035 stars11 Jul 2026: 4,045 stars18 Jul 2026: 4,059 stars25 Jul 2026: 4,068 stars1 Aug 2026: 4,073 stars8 Aug 2026: 4,076 stars15 Aug 2026: 4,085 stars22 Aug 2026: 4,088 stars29 Aug 2026: 4,092 stars5 Sept 2026: 4,094 stars12 Sept 2026: 4,095 stars19 Sept 2026: 4,098 stars26 Sept 2026: 4,100 stars3 Oct 2026: 4,108 stars10 Oct 2026: 4,115 stars17 Oct 2026: 4,120 stars24 Oct 2026: 4,124 stars31 Oct 2026: 4,126 stars7 Nov 2026: 4,129 stars14 Nov 2026: 4,153 stars

3,961 on 25 Apr 20264,153 on 14 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Langroid is a Python library where Agents, each optionally holding an LLM, a vector store and tools, are wrapped in Tasks and solve problems by exchanging messages. It includes DocChatAgent for RAG, SQLChatAgent, function-calling via ToolMessage, and an MCP tool adapter. It works with OpenAI-compatible APIs and local models such as Ollama, and does not depend on LangChain.

Who it is for: Python developers building multi-agent LLM apps with tools and RAG

Strengths

  • Small Agent and Task abstractions; a 2-agent loop takes about 20 lines
  • Supports several vector stores: Qdrant, Milvus, pgvector, Pinecone, Weaviate
  • MCP tool adapter converts server tools into ToolMessage instances
  • No LangChain dependency; parsers and most vector DBs are optional extras

Weaknesses

  • Pull requests accepted by invitation only, tied to real application use
  • OpenAIAssistant class removed in 0.68.0 after the Assistants API sunset
  • Vector-store env-var naming changed in 0.67.0, requiring migration
  • Code-only; no visual builder or hosted deployment layer

What it needs

  • GPU optional
  • Docker
  • Needs LLM provider or local OpenAI-compatible server (e.g. Ollama), Optional vector store (Qdrant, Milvus, pgvector, etc.)
  • Models: OpenAI, Ollama, Gemini, DeepSeek, OpenRouter

What changed

  1. 0.69.24 ↗ (opens in a new tab)Security fix

    “This is a maintenance release: no library code changed, only the locked”

From the release notes. No flag means the note does not say so, not that the upgrade is safe.

Also in Agent frameworks and SDKs

See all 153
Also in Agent frameworks and SDKs
RankProjectScore
1ppt-masterAgent skill that turns documents into natively editable PowerPoint decks Live demo ↗ (opens in a new tab)78k starsMIT96 out of 100
2copilotkitReact and Angular SDK for adding agent chat and generative UI to apps38.8k starsMIT92 out of 100
3openai-agents-pythonPython SDK for multi-agent workflows with handoffs, guardrails and tracing30.7k starsMIT92 out of 100
#4mastraTypeScript framework for building AI agents, workflows and MCP servers29.6k starscustom license92 out of 100
#5browser-usePython library that lets an LLM agent control a web browser Live demo ↗ (opens in a new tab)121.4k starsMIT90 out of 100