TabbyML/tabby
TabbyML
33.5k
Self-hosted AI coding assistant
1.7k|Rust
NOASSERTIONaicodegencoding-assistant
Code
AI coding tools for developers who want local or private assistance near the codebase.
Use this page when code completion, repository chat, terminal pair programming, or model-flexible development workflows need to stay inside your environment.
Repository context
Continue and Aider are useful when an assistant needs to read, explain, or modify code with project context.
Private completion
Tabby fits teams that want self-hosted code completion without sending the developer workflow to an external service.
Why it works
Self-hosted completion
Tabby and Refact are closest when the main need is code completion, IDE assistance, and a private team-controlled backend.
Repo-aware assistance
Continue, Aider, Cline, OpenCode, Codex, Qwen Code, Gemini CLI, and Void work well when an assistant needs to inspect, explain, or modify a repository.
Autonomous coding agents
OpenHands fits heavier autonomous coding work where the agent needs a workspace, shell access, browser tools, or longer task execution under your control.
Curated repositories
AI developer tools
continuedev/continue
continuedev
32.8k
⏩ Source-controlled AI checks, enforceable in CI. Powered by the open-source Continue CLI
4.4k|TypeScript
Apache-2.0open-sourcedeveloper-toolsai
Aider-AI/aider
Aider-AI
44k
aider is AI pair programming in your terminal
4.3k|Python
Apache-2.0chatgptclicommand-line
cline/cline
cline
61k
Autonomous coding agent right in your IDE, capable of creating/editing files, executing commands, using the browser, and more with your permission every step of the way.
6.3k|TypeScript
Apache-2.0Developer Tools & Integrations
anomalyco/opencode
anomalyco
150.2k
The open source coding agent.
17.3k|TypeScript
MITAgentic AI & Multi-Agent Systems
smallcloudai/refact
smallcloudai
3.5k
AI Agent that handles engineering tasks end-to-end: integrates with developers’ tools, plans, executes, and iterates until it achieves a successful result.
309|Rust
BSD-3-Clausefine-tuningself-hosteddeveloper-tools
openai/codex
openai
78.2k
Lightweight coding agent that runs in your terminal
11.2k|Rust
Apache-2.0Developer Tools & Integrations
QwenLM/qwen-code
QwenLM
23.9k
An open-source AI agent that lives in your terminal.
2.3k|TypeScript
Apache-2.0Developer Tools & Integrations
google-gemini/gemini-cli
google-gemini
102.5k
An open-source AI agent that brings the power of Gemini directly into your terminal.
13.3k|TypeScript
Apache-2.0geminigemini-apiai
voideditor/void
voideditor
28.7k
No description provided.
2.5k|TypeScript
Apache-2.0cursoreditorchatgpt
Selection guide
How to choose a private AI coding tool
The main split is server-based team completion, IDE assistants, terminal pair programming, and autonomous coding agents. Each solves a different developer job.
Tabby for self-hosted completion
Best fit when a team wants an on-prem Copilot-style server, IDE extensions, repository context, and admin control.
Aider for terminal pair programming
Best fit when developers want a CLI assistant that can edit multiple files, understand a repo map, and work through git diffs.
Continue for private-model IDE workflows
Useful for IDE chat, edit, and autocomplete with local or private model backends, but weaker as a self-hosted server product.
Agents for multi-step code work
OpenHands, Cline, OpenCode, Codex, Qwen Code, Gemini CLI, and Void fit better when the assistant needs to inspect a repo, run commands, or make multi-file changes.
Privacy fit
Private does not always mean fully self-hosted
Some tools run as local CLI or IDE clients that connect to private model endpoints. Others run a central server. Both can improve privacy, but they are different deployment models.
Centralized teams
Tabby and Refact are stronger matches when teams need central deployment, code-completion control, or a private assistant backend.
Individual developers
Aider, Continue, Cline, OpenCode, Codex, Qwen Code, Gemini CLI, and Void can be strong fits when a developer controls the model endpoint and workflow locally.
Workflow fit
Autocomplete, chat, or code changes
Rankable comparison content should explain whether a project is built for inline completion, repository chat, AI checks, terminal pair programming, or autonomous multi-file changes.
Do not flatten the category
Tabby, Continue, Aider, OpenHands, Cline, and OpenCode all help developers, but they should not be presented as interchangeable Copilot clones.
Suggested additions
Strong candidates not yet in the registry
Roo Code
RooCodeInc/Roo-Code
An open-source VS Code coding agent with local and OpenAI-compatible model support. Strong fit, but suggested until the canonical repo is present upstream.
View repositorymini-SWE-agent
SWE-agent/mini-swe-agent
A lightweight coding agent for repo and issue-fixing workflows. Useful for agentic coding experiments, but narrower than IDE assistants and terminal pair programmers.
View repositoryPlandex
plandex-ai/plandex
A terminal-based coding agent designed for larger tasks and multi-file plans. Relevant, but activity and fit should be reviewed before promoting it to the main list.
View repositoryTrae Agent
bytedance/trae-agent
A software-engineering agent for coding tasks. Strong enough to watch, but less established in self-hosted developer workflows than the main curated tools.
View repositoryKilo Code
Kilo-Org/kilocode
An open-source agentic engineering platform and IDE assistant. Relevant to coding-agent intent, but should stay suggested until registry/source backing is clear.
View repositoryOpen SWE
langchain-ai/open-swe
An asynchronous coding agent from the LangChain ecosystem. Useful for background code tasks, but newer and more specialized than the main list.
View repositoryDeepCode
HKUDS/DeepCode
An agentic coding project for paper-to-code, text-to-web, and backend generation. Interesting but more specialized than general private coding assistants.
View repositoryOnlook
onlook-dev/onlook
An open-source AI-first design and React editing tool. Adjacent to developer tools, but stronger for visual frontend/design workflows than general coding assistance.
View repositoryRelated pages
Keep browsing
Self-hosted ChatGPT alternatives
Chat interfaces and assistant apps you can run with local models, private endpoints, or your own hosted providers.
Local model runtimes and inference servers
Inference servers and local runtimes for serving models on your machine, server, or private cloud.
Self-hosted RAG tools
Knowledge-base apps, retrieval frameworks, and document pipelines for private data and production AI systems.
Vector databases and retrieval storage
Databases and search layers for embeddings, metadata filtering, persistence, and semantic retrieval.
Agents, workflows, and app builders
Agent frameworks, workflow engines, and app builders for repeatable AI-powered processes.
FAQ
Questions answered
What counts as an AI developer tool?
A self-hostable tool that helps developers write, review, understand, or navigate code while keeping the workflow private.
Should coding assistants live on the chat page?
No. Coding assistants have a different job: code completion, repository context, terminal work, and development workflows rather than general assistant chat.
What is the best self-hosted AI coding assistant for teams?
Tabby is the strongest fit when teams need a central self-hosted coding assistant with IDE extensions, repository context, and admin control. Refact is another team-oriented option. Continue, Aider, Cline, OpenCode, Codex, Qwen Code, Gemini CLI, and Void are better when individual developers control their local workflow and model endpoint.
How are Aider and Continue different from Tabby?
Aider is a terminal pair-programming tool that edits code through git-aware workflows. Continue is an IDE assistant that can use private or local model backends. Tabby is closer to a self-hosted server product for team code completion.
When should I use an autonomous coding agent?
Use OpenHands, Cline, OpenCode, Codex, Qwen Code, Gemini CLI, or similar tools when the job needs multi-file edits, shell commands, repository inspection, or longer task execution. Use Tabby or Continue when the job is mostly completion, chat, and inline editing.