Eclipse Theia
An extensible open-source framework for building your own cloud and desktop IDEs with web technologies — with an AI layer that stays vendor-neutral.
- Source repo
- eclipse-theia/theia
- Stars
- ★ 22k
- Last updated
- today
- License
- EPL-2.0
- Primary language
- TypeScript
- FA score
- 49/100 · Major gaps
At a glance
- How it runs
- Works with
- Portable with changes
- Cost
- Free, no paid service needed
- Setup effort
- High · needs real infrastructure
- You'll need
- Typical use
- A tool vendor wants to ship a branded, customized IDE without reimplementing the editor, terminal and language services.
- Not a fit if
- Individuals who want a ready-made IDE with no coding
- Teams expecting a vendor-hosted cloud IDE service
- Groups unwilling to maintain a Node.js/Yarn build and custom extensions
- Website
- theia-ide.org ↗ set on GitHub
- Maintenance
- Commits in the last 30 dayslast commit 2026-11-20latest release v1.77.0 (2026-11-05)+3.7 ★/day over 7 days
- Source review
- 49/100 · Major gaps
What does this agent do, and when should you use it?
Eclipse Theia is an extensible framework for building full-featured, multi-language cloud and desktop IDEs and tools; it is not a finished IDE but a platform adopters compose their own product on. It is implemented in TypeScript on modern web technologies and can run in the browser or be packaged as a desktop application with Electron. The framework supports the VS Code Extension protocol, so extensions from the VS Code ecosystem can run inside Theia. The project also ships Theia AI, described as a vendor-neutral AI harness platform, which powers the AI-native Theia IDE. Theia is governed vendor-neutrally under the Eclipse Foundation and licensed under EPL-2.0.
Theia supplies the shell and services of an IDE: editor, file explorer, terminal and language services are assembled from composable extension points, and adopters add custom views, commands and backend logic by writing Theia extensions. The framework implements the VS Code Extension protocol and Language Server Protocol so existing VS Code extensions and language servers can run, and it supports the Model Context Protocol (MCP) as the channel through which Theia AI reaches external tools and context services. Theia AI acts as the harness connecting LLM capabilities to IDE interactions. The output is a browser application or an Electron desktop application that you build yourself — the packaged Theia IDE is produced in the theia-blueprint repository.
- A tool vendor wants to ship a branded, customized IDE without reimplementing the editor, terminal and language services.
- An enterprise wants its internal build, deploy and review tooling surfaced inside a browser-accessible development environment for staff.
- A team wants to reuse the VS Code extension ecosystem without being tied to VS Code's own product shape and governance.
- A team embedding LLM assistance into an IDE while avoiding lock-in to a single model provider.
- An adopter that needs a cross-platform Electron desktop IDE while sharing one codebase with a web frontend.
- Research and teaching settings where AI harness and MCP tool integrations need to be experimented with inside an IDE.
How do you install or deploy this agent?
The README points to the project website's getting-started guide but does not include the concrete install commands in this repository. The known runtime prerequisites are a Node.js and Yarn build chain (exact versions are not stated in this source). The typical path is running the examples from source:
How do you use this agent?
Per the contributing section, fork and clone the repository, then follow the quick start in doc/Developing.md to run the examples from source. To build your own application, the website's "Composing Applications" and "Authoring Extensions" guides cover the steps. This README page does not spell out a complete copy-paste command sequence, so consult those documents for exact commands.
What are this agent's strengths and limitations?
- Framework-first design: adopters compose a branded IDE of their own rather than configuring a fixed product.
- Implements the VS Code Extension protocol, letting existing VS Code extensions be reused.
- One TypeScript codebase targets both browser-based and Electron desktop delivery.
- Vendor-neutral open-source governance under the Eclipse Foundation with an EPL-2.0 license.
- Ships Theia AI as a vendor-neutral AI harness and supports MCP as a single integration surface for AI in the IDE.
- Produces an SBOM for every release and publishes it to the Eclipse Foundation SBOM registry, aiding supply-chain transparency.
- It is a framework, not a ready IDE: you must write extensions and maintain them yourself.
- Large, modular architecture with a steep learning curve for first-time adopters.
- No runnable install or build commands are given on this page; you must go to the website and doc/Developing.md.
- Desktop delivery depends on Electron, and browser delivery requires you to design and operate your own hosting.
- The README says little about Theia AI's internals or which model providers it supports, making the AI boundary hard to assess from this source.
How does this agent compare with similar options?
The README names no competitors; its VS Code Extension protocol support implies a relationship to the VS Code ecosystem, but the source makes no direct comparison.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Eclipse Theia This agent | 49 · Major gaps | FrameworkFree | ★ 22k | today | TypeScript | — |
| FastMCP | 68 · Some gaps | Library / SDKFreemium | ★ 29k | today | Python | — |
| Serena | 74 · Some gaps | CLIFreemium | ★ 32k | 1d ago | Python | Codex · Claude Code · Claude.ai |
| Crush Terminal Coding Assistant | 65 · Some gaps | CLIFree + model costs | ★ 29k | today | Go | OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows SECURITY.md coordinated disclosure, extensive package.json overrides and an allowScripts allowlist, and README mentions SBOM, indicating some supply-chain governance; however, as an IDE/AI platform, AGENTS.md and CLAUDE.md only describe development conventions and give no runtime policy for least privilege, user confirmation, data flow, or sensitive data handling for AI agents, so those score 1. Dependency security scores 2 for explicit overrides/allowScripts but no visible vulnerability scan results. Rollback only appears via CHANGELOG/migration hints, no runtime rollback, score 1. Source attribution has SPDX headers and trademark notice, score 2.
AGENTS.md and CLAUDE.md consistently describe build/test/start commands (npm, build:browser, esbuild), self-consistency 2; dependency availability only shows engines node>=24 and npm ci with no offline/mirror guidance, score 1; failure messages are barely visible in the sampled source (only mock client throws), score 1.
README clearly targets browser/desktop IDE builders and VS Code extension authors, score 2; capability boundaries appear in Scope and AGENTS.md FileService restricted-path notes, score 2; trigger precision is weak because the skill description is broad ('update default models') with no precise trigger conditions, score 1; environment fit is supported by Node 24, Electron, and multi-platform CI matrix, score 2.
README has clear TOC and getting-started links, information architecture and install notes each 2; naming stability (kebab-case, package naming) 2; examples and FAQ are mostly external links with limited in-repo samples, score 1; known limitations are scattered (webpack removal, FileService limits) rather than consolidated, score 1; dual license with full SPDX headers, score 3; versioning and CHANGELOG policy explicit, score 2; maintenance responsibility points to Eclipse Foundation and release process but publisher identity is unverified, score 2.
Output usability: docs and skill steps are concrete and actionable, score 2; marginal value as a mature IDE framework is clear, score 2; cost-benefit is high for ordinary adopters given the large monorepo and long builds (60-minute CI timeouts, multi-platform matrix), score 1.
README and AGENTS.md claims mostly link out to website/wiki, limiting in-repo traceability, score 1; cross-source corroboration between CI workflows and package.json scripts is partial, score 1; fact/inference separation is not explicitly marked in the prose, score 1.
- AGENTS.md/CLAUDE.md are contributor development guides, not runtime AI-agent safety policies; do not infer least privilege or user confirmation from them.
- AI packages (ai-core, ai-openai, etc.) maintain model defaults and capability tables manually via a skill, risking drift from provider APIs.
- The repository is large with long build times; assess CI and local resource cost before adoption.
- Publisher identity is unverified by FollowAgents; maintenance responsibility and update path should be independently confirmed.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
Automated next build of native dependencies from master (commit 727005ec31a7d0a6ac5da4877fbaf66bbacf9aef, 2026-11-20 00:41 UTC).View on GitHub ↗ -
v1.77.0 Eclipse Theia v1.77.0
Release notes excerpt
What's Changed - [Release Notes](https://github.com/eclipse-theia/theia/blob/release/1.77.x/CHANGELOG.md#1770---1152026) - [Breaking Changes](https://github.com/eclipse-theia/theia/blob/release/1.77.x/CHANGELOG.md#breaking_changes_1.77.0) Contributors @Alexander-Taran, @AmanDeka, @bates64, @CamilleLetavernier, @cdamus, @colin-grant-work, @daviwil, @dr14-make, @EhabY, @eneufeld, @eng-abdelrahman-sharaf, @fismif, @gastonfeng, @j1147v, @jfaltermeier, @JonasHelming, @kachurun, @kaikongbj, @KevinGruber2001, @LordBaryhobal, @lucas-koehler, @MatthewKhouzam, @mehedi-107, @ndoschek, @richter-m …View on GitHub ↗ -
v1.74.3 Eclipse Theia v1.74.3
Release notes excerpt
Based on https://github.com/eclipse-theia/theia/tree/v1.74.2 Includes the following fixes: * aa5de6438acacf2223b27235fc9a08c9c573a789: chore(electron): upgrade Electron from 42.11.8 to 42.11.16 Full Changelog: https://github.com/eclipse-theia/theia/compare/v1.74.2...v1.74.3View on GitHub ↗
FAQ
Can I just download and use Theia as an IDE?
Is it free?
Will my existing VS Code extensions run on Theia?
What do I need to get started?
Does Theia AI lock me into one model provider?
Is Eclipse Theia open source?
Is Eclipse Theia safe to use?
Is Eclipse Theia still maintained?
Who makes Eclipse Theia?
What is Eclipse Theia's official website?
Thanks — we use this to improve the page.