克隆必须带 --recurse-submodules，否则子模块缺失；源码构建依赖 make 与 Python 3.10–3.13；make flush 会删除 ~/.anton 中的密钥和 ~/.cowork 中的对话与数据库
mindshub
mindsdbMIT活跃维护把开源模型和 Agent 统一到一个工作空间里，让 Claude、DeepSeek、GLM 等模型直接帮你干活，省去自己拼装编排的麻烦。
The unified workspace where open-source models get things done for you.
v26.1.0发布于 2026-04-23 07:11 UTC39,852star +6 / 7 天6,229fork同步于 2026-11-10 11:20 UTC
git clone --recurse-submodules https://github.com/mindsdb/minds.git
cd minds
make setup
make dev-web 硬件要求按 README 说明
README 未说明显存或 GPU 要求；桌面端提供 macOS 与 Windows 安装包，Linux 需从源码构建，也可直接用网页版免安装
开源智能体工作台，用可切换的开源 harness 和自选模型完成知识工作与软件开发
把开源模型和 Agent 统一到一个工作空间里，让 Claude、DeepSeek、GLM 等模型直接帮你干活，省去自己拼装编排的麻烦。
核心能力
- 用 Model Router 在 Claude、GPT、Gemini 与 DeepSeek、Qwen 等模型间切换
- 运行 Anton、Hermes 等可互换的开源 agent harness
- 通过数据保险库连接 BigQuery、Postgres、Gmail 等系统且凭据按连接隔离
- 把 agent 输出转成文档、仪表盘、应用并发布到在线 URL
- 提供跨会话记忆、可复用技能库与定时任务调度
容易踩的坑
内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README
- v26.1.0最新正式版2026-04-23 · 2026-04-23 07:11 UTC
TL;DR
Changes to SQL
There have been several notable updates improving SQL support: Additional MySQL API tests, fixes for MSSQL Query Execution Result Conversion, improved scoping on MySQL's get_columns schema and parsing on BigQuery's Select response. Support for additional gitlab configurations has been added. To ensure failsafe operation, proper exception handling has been put in place for MSSQL check_connection and error handling mechanisms are implemented for temporary folder cleanup.
Changes to Integrations
A good number of new integration handlers have been added this time including: Leads to Hubspot Handler, Raindrop.io handler, Denodo Integration handler, and Freshdesk integration. Fixes and updates have been made to Hubspot handler and support has been added for additional gitlab configurations, Motherduck is now supported in DuckDB, Azure provider was fixed. Various fixes have been made to the boot error messages in integration handlers.
Changes to Knowledge Bases
There have been two important fixes to Knowledge Bases. One ensuring the creation of knowledge base without explicitly providing provider configurations is successful for non-openAI scenarios. The second optimizes get_integration by fetching directly from DB.
Bug Fixes and Improvements
There have been improvements to error and crash handling, along with particular issues with Duplicated columns, PID file creation, and IN filter with the API handler. There has been optimization in mixed search with improvements to the faiss ivf file and allowing data handlers to stream data.
Changes to Documentation
Various documentations have been updated including ENV Variables, datastax.mdx, and gitlab handler readme. There have been corrections to CREATE MODEL/PREDICTOR typo on model creation page and updates to the home page have been done.
What's Changed
[... The entire "What's Changed" section you provided is pasted here, without any changes ...]
- v26.1.0rc1pre-release2026-04-17 · 2026-04-17 13:36 UTC
查看更新说明
TL;DR
Changes to SQL
- Additional MysQL API tests have been added for expanded coverage, enhancing reliability.
- Improved MSSQL query execution result conversion and schema scoping.
- Enabled support for additional gitlab configurations.
- MariaDB tests have been fixed.
- MySQL got_columns schema scoping was updated for better compatibility.
- BigQuery newly added parameters result in better control and flexibility.
Changes to Integrations
- Documentation for datastax has been updated, improving clarity and usability.
- Improved the CRUD operations of hubspot handler.
- Added Leads to Hubspot handler for more functionality.
- Introduced the ability for data handlers to stream data, enhancing data handling efficiency.
- Fix for queries to API handler with
INfilter, providing better consistency. - Optimal get_integration by fetching directly from DB improved performance.
- Improved support for additional gitlab configurations.
- ClickHouse data catalog was added, expanding data source options.
- Raindrop.io integration handler was added with full CRUD support and complete test coverage.
- Denodo integration handler was added, expanding data source options.
- Freshdesk Integration was added, broadening third-party tool integration options.
- Jira Handler updates improve functionality.
Changes to Knowledge Bases
- Optimization for the creation of knowledge bases for non-OpenAI scenarios was enhanced.
- Temporary folder cleanup error handling was improved, enhancing system stability.
Bug Fixes and Improvements
- Fixed various issues with handlers for Hubspot, Hana, Confluence, and others. These fixes have improved the reliability and functionality of the handlers.
- The consistency of functionalities like lock the PID file or support of xls files has been improved with the application of respective fixes.
- Static errors and issues with subquery references were resolved for improved stability.
- Specific security enhancements added through updates to protobuf, python-multipart, socket.io-parser, requests, yaml, brace-expansion, basic-ftp, and filelock, to bolster security.
- Motheduck support was added to DuckDB for optimization.
- Several fixes applied to the formed JSON exception type in Postgres to increase stability.
Changes to Documentation
- Various documentation updates including ENV variables, gitlab, hubspot and others for better understanding.
- A mintlify update improved the aesthetics of the code.
- The addition of Faiss doc improved clarification on the utilization of Faiss.
What's Changed
[The What's Changed section was not modified as per instructions. Refer to the input.]
Full Changelog: https://github.com/mindsdb/mindsdb/compare/v26.0.1...v26.1.0rc1
- v26.0.12026-03-03 · 2026-03-03 16:02 UTC
查看更新说明
TL;DR
Changes to Documentation
The
README.mddocument has been updated. The document now contains the latest instructions and details for users, which will aid in precise understanding and improved user experience.Changes to SQL
We have introduced a native query passthrough. This feature allows the system to understand and directly execute native queries which leads to significant enhancements in the efficiency of query execution.
What's Changed
- Update README.md by @C0staTin in https://github.com/mindsdb/mindsdb/pull/12248
- have native query passthrough by @torrmal in https://github.com/mindsdb/mindsdb/pull/12264
Full Changelog: https://github.com/mindsdb/mindsdb/compare/v26.0.0...v26.0.1
- v26.0.02026-02-25 · 2026-02-25 10:25 UTC
查看更新说明
TL;DR
Changes to SQL
The update includes several improvements to our SQL capabilities. We've made a fix for
GROUP BY WITH ROLLUP, making it more reliable and robust. TheMINDSDB_DB_CONenvironmental variable now incorporates validation to minimize potential errors. We've made thetrack_columnparameter case-insensitive, thus enhancing its usability. Our SQL parsing has been improved to prune ambiguous columns and extract query targets in the API Handler accurately. A significant development ensures proper handling and utilization of memory when DuckDB is used. Also, the available memory check in the Snowflake handler has been updated and improved.Changes to Integrations
There have been significant enhancements to our integrations. README's for integrations are now pulled statically from GitHub, thereby reducing the need for local storage. The Shopify, Confluence, Databrick, Hubspot, and Netsuite handlers have been updated and improved for better interactions. Additional validation is now in place for the targets list in the Shopify handler. We have also deprecated Dspy, Chromedb, and all ML handlers for a cleaner, more streamlined codebase. Images have been added for the readme files to enhance understanding of the integration process.
Changes to Knowledge Bases
We've made some notable changes to our Knowledge Bases. The default Knowledge Base store in Docker-compose has been switched to pgvector for enhanced performance. Display of mixed case columns in the Knowledge Base has been fixed. Fixes have been provided for Knowledge Base ID duplicates on the second insert and for pgvector. Batches are now the default method for inserts in the Knowledge Base.
Bug Fixes and Improvements
Several bug fixes and improvements have been rolled out. These include rectifications in README locations, language permission issues, Shopify handler query limit, and more. There have also been fixes for reported bugs such as file datasource visibility post Docker Upgrade and Query charting failure among others.
Changes to Documentation
We've performed a clean-up of the codebase, deleting unnecessary files for better readability and understanding.
Security Upgrades
Numerous security upgrades have been implemented covering packages like urllib3, starlette, keras, protobuf, numpy, aiohttp, fugue, sqlparse, filelock, azure-core, pyasn1, mintlify, sentencepiece, marshmallow and others to ensure the safety and reliability of the system.
Attention! The release contains breaking changes.
New Contributors
- Welcome to our new contributors: @PPeitsch and @Sweetdevil144
Full Changelog: https://github.com/mindsdb/mindsdb/compare/v25.14.1...v26.0.0
- v26.0.0rc3pre-release2026-02-20 · 2026-02-20 08:33 UTC
查看更新说明
TL;DR
Changes to Integrations
Our team enabled Netusite cloud which brings with it a host of improvements, considerations and necessities. By providing support for Netusite cloud, our users will be able to access MindsDB capabilities on this platform. This is a significant leap towards supporting a broader range of cloud platforms enabling users choose the platform they are most comfortable with.
What's Changed
- Enable netusite cloud by @tino097 in https://github.com/mindsdb/mindsdb/pull/12235
Full Changelog: https://github.com/mindsdb/mindsdb/compare/v26.0.0rc2...v26.0.0rc3
用户点评
登录用户的真实使用记录，可按版本和硬件筛选共 0 条
还没有人点评。第一个说出真实体验的人最有价值。
MindsHub is an agent workspace for completing knowledge work and developing software, with open-source agent harnesses and a choice of models.
Website · Docs · Web app · Pricing · Discord
MindsHub gives you the freedom to choose among models and providers instead of being locked into one ecosystem. Use its agent workspace for research, analysis, content creation, and software development. It is open source and can run on your machine, in your VPC, or through the hosted app.
This repository is the platform superproject: it pulls together the desktop/web app, the agent backend, and the data engine so you can build and run the whole stack from source.
For building intelligence into agents and products through one API, see MindsHub Inference. It provides a choice of models and providers with one set of controls and one bill.
Get started
Pick whichever fits:
- Web — nothing to install. Open console.mindshub.ai and sign in.
- macOS. Download the desktop app (
.pkg). - Windows. Download the desktop app (
.exe). - Linux. Build from source.
Free to start. Pro adds all frontier models and private artifacts — see pricing.
What you can do
For every knowledge worker — creators, strategists, and operators:
- Automate repetitive, multi-step work that involves reading and writing: reports, monitoring, recurring workflows, and scheduled operations.
- Build internal AI tools and artifacts — apps, dashboards, decks, docs, analyses — without engineering, and publish them to a live URL to share with your team.
What's inside
- Connected data. A secure vault links systems like BigQuery, Postgres, Gmail, Drive, HubSpot, Notion, and Linear. Credentials stay scoped per connection — agents never see raw keys.
- Model Router. Switch between frontier models (Claude, GPT, Gemini) and open models (DeepSeek, Qwen, Kimi) without wiring up a key for each provider.
- Open agents. Run interchangeable open-source harnesses — Anton (default) and Hermes — swappable from a dropdown.
- Artifacts. Turn agent output into documents, dashboards, apps, and code, and publish to a live URL.
- Memory, skills & scheduling. Cross-session memory, a reusable skill library, and tasks that run on a schedule.
Build from source
1. Clone the repository
git clone --recurse-submodules https://github.com/mindsdb/minds.git
cd minds
2. Install dependencies
make setup
3. Run
| Mode | Command |
|---|---|
| Desktop app (Electron) with hot reload | make dev or make watch |
| Web app in browser with hot reload | make dev-web |
| Production build | make build |
| Package for macOS | make dist-mac |
| Package for Windows | make dist-win |
Build macOS .app from local uncommitted source | make pack-local |
| Wipe all local installs + data (fresh start) | make flush |
Fresh start:
make flushremoves the local runtime (thecowork-serveruv tool and thebackend/*/.venvs) and deletes app state in~/.anton(provider keys) and~/.cowork(database, hermes, projects). Use it to test the from-scratch install flow or recover from a broken install. It prompts for confirmation — passFORCE=1to skip. The nextmake setupor app launch reinstalls everything. ⚠️ This deletes your conversations and saved keys.
Working on feature branches (submodules)
This repo is a superproject that pins each module (frontend, backend/core_api, backend/core_agent, backend/data-vault) to a commit. To work on module branches without polluting git status or fighting over pins:
1. Pick your branches in a gitignored dev.env (copy the template):
cp dev.env.example dev.env # then set REF=feat/my-thing (or per-module API_REF=…)
2. make follows it — one knob, both run paths:
| Command | What it does |
|---|---|
make use | check out your dev.env refs across all submodules |
make dev / make watch | run the Electron app with live reload against local source |
make dev-web | run the web SPA with live reload against local source |
make server + make app | (re)install the desktop server from the configured branch, then launch |
make server-local + make app-local | install the desktop server from local uncommitted source, then launch |
make pack-local | build the macOS .app from local uncommitted source (no push needed) |
make refs | show which refs the next run will use |
make baseline | reset submodules to the pinned commits |
make pin | record the current submodule commits as the superproject's pins (one deliberate commit) |
Submodules are configured with ignore = all, so your branch work never shows up as superproject changes — the parent git status stays clean. Pins move only via make pin. See CLAUDE.md for the full workflow.
Deploy anywhere
MindsHub is built for flexible deployment — cloud, VPC, on-prem, air-gapped, and hybrid infrastructure — so you keep full control over your infrastructure, models, permissions, and data.
Help & support
- Ask a question — join the Discord community.
- Report a bug — open a GitHub issue with reproduction steps.
- Read the docs — guides, setup, and the API at docs.mindshub.ai.
- Enterprise SLAs or custom deployments — contact the team.
🤝 Contribute
MindsHub is open source and contributions are welcome — code, integrations, docs, bug reports, and feature ideas. Read the docs to get set up, browse open issues, and say hi on Discord.
🔒 Security
Found a security vulnerability? Please don't open a public issue. Report it privately through our security policy.
📚 Resources
📄 License
This repository is released under the MIT License. Bundled components are governed by their own licenses — see each submodule's repository for details.