other / agent

LibreChat

A self-hosted AI chat platform with agents, MCP support, skills, artifacts and a code interpreter for many model providers.

Repository owner LibreChat-AI

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: A self-hosted AI chat platform with agents, MCP support, skills, artifacts and a code interpreter for many model providers.
Runs in: Web service
Operating systems: Not stated
Needs: Runtime: Not stated; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
Tool type: Agent
Delivery: From GitHub
Version: Latest on the main branch
Access: Code on GitHub
Compatibility: According to the README, LibreChat connects to Anthropic (Claude), AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI and the OpenAI Responses API. The README says custom endpoints support any OpenAI-compatible API and lists local and remote providers such as Ollama, AMD Lemonade, Groq, Cohere, Mistral AI, Apple MLX, koboldcpp and OpenRouter. It lists MCP support, sign-in with OAuth2, LDAP and email, and speech features that support OpenAI, Azure OpenAI and Elevenlabs. The README lists a multilingual interface.
External costs: The README names commercial model providers and other services, including OpenRouter, Perplexity, Elevenlabs and Jina reranking, alongside local providers such as Ollama. The README also links cloud deployment services Railway, Zeabur and Sealos; its Railway button uses a referral link. The README gives no prices. The repository is MIT-licensed. Your coding agent or model provider may charge for usage.
Permissions: The README offers cloud deployment buttons and describes Docker Compose stacks for installation. The README says agents can use MCP servers, file search, code execution and web search. The README describes highly experimental attached workspaces where agents inspect, read, search and edit files and run Bash with bounded timeouts. The README describes Ask, Allow and Deny controls for file writes and command execution where administrators permit them, including Full access mode for attached environments. The README says the beta Agent Management API can create, update and delete Agents and manage Agent files and Skills. The README describes an administration panel for managing users, groups, roles and configuration overrides and editing settings and permissions live. The README also describes always-on Skill workflows, background tools and experimental startup-loaded Agent Plugins. The top-level entries include scripts/; the README does not describe that folder's behaviour.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms.
Interface: web

Code on GitHub

LibreChat-AI/LibreChat

MIT

TypeScript45,598 stars9,373 forksLast pushed 13 Oct 2026 UTC

Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, Secure Multi-User A

Topics: ai anthropic artifacts aws azure chatgpt chatgpt-clone claude

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

View on GitHub ↗

What it does

The README describes LibreChat as a self-hosted AI chat platform with an interface inspired by ChatGPT. The README lists Anthropic, AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI and the OpenAI Responses API, alongside custom OpenAI-compatible endpoints and local and remote providers.

The README lists no-code agents, sharing with users and groups, MCP support, SKILL.md instruction bundles, subagents, a code interpreter, web search, code artifacts, image generation, presets and conversation search. The README also lists OAuth2, LDAP and email sign-in and a browser-based administration panel.

The README's version 0.8.8 section labels the Agent Management API beta and attached workspaces highly experimental. The README labels Agent Plugins experimental.

Who it is for

People and teams who want to host their own chat interface for many AI models, with agents and multi-user sign-in.

Example tasks

Switch between AI endpoints and presets in the middle of a chat.
Build a no-code custom agent and share it with specific users and groups.
Create a reusable Skill from a SKILL.md instruction bundle.
Connect an MCP server to give an agent tools.
Import conversations from ChatGPT and export conversations as markdown or JSON.

Getting started

The README gives no install commands. The README points to its documentation and shows deployment buttons for Railway, Zeabur and Sealos; the Railway button uses a referral link. The README says the administration panel is bundled with the Docker Compose stacks for one-command setup. The repository's top-level entries include docker-compose.yml, Dockerfile and helm/. Follow the documentation linked by the README for the full steps.

Limitations

According to the README, LibreChat is self-hosted. The README gives no install commands. It points to its documentation and to deploy buttons for Railway, Zeabur and Sealos. It tells readers to check the changelog for breaking changes before updating. It marks the Agent Management API as beta, attached workspaces as highly experimental and Agent Plugins as experimental. It says the code interpreter is powered by ClickHouse/code-interpreter and that resumable streams in horizontally scaled deployments use Redis. The repository can change at any time; check its README before you install.

Data practices

The README describes LibreChat as self-hosted and says you can use it completely local or deploy it on the cloud. Prompts and files go to the model providers and endpoints you configure, and web search uses the search providers, content scrapers and rerankers you configure. The README lists a Trace Viewer for model conversations, tool rounds and costs, export of traces and application logs through OpenTelemetry, and a Langfuse connection with allowlisted trace identity and metadata. The README also describes browser diagnostics tagged with client build IDs, Insights scoped to authorized Agents, and Context Usage displays. The README does not describe an opt-out for these features.