Eclipse Theia 1.61 Release: News & Noteworthy

We are happy to announce the Eclipse Theia 1.61 release! The release contains in total 52 merged pull requests. In this article, we will highlight some selected improvements and provide an overview of the latest news around Theia.

For those new to Eclipse Theia, it is the next-generation platform for building IDEs and tools for the web or desktop, based on modern state-of-the-art web technologies. With Theia AI, part of the Theia platform, you can also build AI-powered tools and IDEs with ease. For more details about Eclipse Theia, please refer to this article and visit the Theia website.

The Theia 1.61 release is the first release candidate for the next community release 2024-05. So if you plan to adopt the next community release, please verify the 1.61 release and report important issues for your use case!

Please note that with the 1.61 release, we have finally removed the deprecated Theia Git extension from the release. If you need Git support for a tool based on Theia, we recommend using the built-in Git VS Code extension, also see this issue.

The Theia project also releases a product, the Theia IDE. The Eclipse Theia IDE is a modern, AI-powered, and open IDE for cloud and desktop environments, aimed at end users. The Theia IDE is based on the Theia platform and also includes advanced AI powered features. For more details, see the Theia IDE website.

If you are looking for a simple way to check out the new release, please download and install the Theia IDE, which is based on Theia 1.61.

Eclipse Theia 1.61: Selected features and improvements

In the following, we will highlight some selected improvements in the new release. As usual, we cannot mention all 52 improvements, however we will focus on the most notable changes as well as changes visible to end users. The corresponding pull requests are linked under the respective heading when applicable.

Theia AI Agent follow-up suggestions

Theia 1.61 introduces support for chat suggestions, a new feature in Theia AI designed to enhance user interaction within the chat interface. Chat suggestions offer users contextual guidance, helping them decide on next steps, access common actions quickly, or receive situational tips during an ongoing conversation.

Suggestions are presented directly in the chat UI (see the screenshot below) and can be augmented to allow the user triggering either Theia commands or custom callback functions when clicked. This enables developers to guide users toward relevant workflows or tool-assisted actions without requiring manual input. In the AI-powered Theia IDE, this feature is already leveraged by the Architect Agent. For instance, the Architect agent prompts users to summarize a task to hand it over to Theia Coder for implementation, as shown in the following screenshot.

For details on how to implement and customize chat suggestions, see the documentation on Chat Suggestions.

Chat Summary and Task Contexts

With Theia AI 1.61, we introduce Task Contexts—a new mechanism for externalizing all relevant information around a specific task to be carried out with the help of AI. A Task Context is a persisted file that stores essential information about a specific task for structured collaboration with agents such as Theia Coder.

Task Contexts can be created either manually or with support from the AI itself. For example, users can provide a feature request or prompt to the Architect agent, who in turn generates a structured summary of the task (see the screenshot in the previous section for a demonstration). When such a summary is confirmed and finalized, a corresponding Task Context is created. This context contains architectural decisions, references to relevant files, and detailed implementation steps—providing a blueprint for downstream agents such as Theia Coder.

Once a Task Context is added to a session, it scopes the conversation to the task at hand. This means agents will focus only on information described within the Task Context, keeping the interaction precise and aligned with the user’s intent. The following screenshot shows an example usage with the Theia Coder agent. We manually add an existing Task Context to a session. Then, we only have to prompt “implement this” to start the workflow, as all relevant information is already captured in the underlying file.

The concept of Task Contexts offers several key advantages:

• Persistent AI memory: A Task Context can be inspected, updated, or corrected by the user, allowing for efficient AI operation while preserving full transparency and control.
• Separation of task phases: Users can independently address different stages of a task—from planning and specification to implementation, testing, and even pull request creation. This structured lifecycle enhances modularity in development.
• Decoupled chat and context: As chat sessions grow long or shift focus, users can offload accumulated context into a Task Context file and start a fresh session without losing continuity. This helps maintain clarity and ensures that important context remains reusable across sessions. You can also share task context easily with your colleagues.

We will publish a dedicated article and tutorial on the usage of Task Contexts in the AI-powered Theia IDE soon, follow us on LinkedIn to get notified!

Add project-info prompt fragment to Theia Coder and Architect

A new default prompt fragment prompt:project-info has been added by default to Theia Coder and Architect. This fragment is designed to provide AI agents with project-specific information that helps them better align with a project’s conventions and architecture. The information is sourced from a project-info file, e.g. stored in a .prompts directory within the workspace. This setup ensures that relevant knowledge about the project is immediately available to the agent as part of its system prompt when the project is opened in Theia.

This enhancement addresses a key challenge in AI-assisted development: agents often miss or misinterpret project-specific conventions, leading to suboptimal suggestions. Instead of relying solely on agent inference or heavyweight solutions like fine-tuning or RAG, the project-info file provides lightweight, explicit guidance — similar to handing a new developer a brief style guide or a best-practice example.

For more details, including a demo and best practices on creating and maintaining effective project-info files, refer to the blog post:

👉 Enhancing AI Coding Agents with Project-Specific Information

We of course also added a project-info file for the Theia repository itself!

Theia AI: Editable Chat Requests

Theia AI 1.61 now supports editing user messages directly within the chat UI. This feature allows users to modify a previously sent request, for example to fix typos or adjust the prompt, without starting over or losing context. Once a message is edited, a new conversation branch is created. Users can then switch between different versions of the request, enabling exploration of alternative responses or iterations of a prompt.

In the updated interface, an edit icon is shown next to each user request. Clicking the icon opens the original message for editing. After changes are submitted, the conversation continues based on the new message, while still retaining access to the original version. This makes it easier to refine prompts, compare different outcomes, or correct earlier mistakes during a session.

The linked video demonstrates a use case where a type is introduced in a request but the language specification (TypeScript) is omitted. By editing the message and restarting the generation, the desired result is achieved, without abandoning the ongoing conversation.

Add support for tracking token usage

Theia 1.61 introduces support for tracking token usage across integrated LLM providers. Token reporting is now implemented for Anthropic, OpenAI, and Google models, allowing users to monitor consumption within the IDE.

A new tab has been added to the AI Configuration view to display token usage on a per-model basis. This view helps developers better understand and manage their token consumption when working with different models. As Theia AI is compatible with a wide range of LLMs (more on this), the new Token view ensures that developers retain visibility and control over resource usage, regardless of the underlying provider.

The following screenshot shows the Token Usage View in action:

Allow workspace specific custom agents

Theia now supports workspace-specific custom agents, allowing to define project-specific AI workflows. When adding a new custom agent in the AI Configuration view, users can now choose between using the global prompt directory or a prompt directory defined at the workspace level. If workspace-specific locations are configured, these take precedence over global configurations—ensuring that agents relevant to the active workspace are prioritized.

Custom agents in Theia allow users to define tailored chat-based assistants that automate specific development tasks directly within the IDE. These agents can access workspace files, variables, and tools , enabling custom capabilities such as test generation, documentation, or security scanning. They are fully integrated into the Theia environment and are defined through simple YAML configurations.

👉 Read the documentation on custom agents for setup instructions and example use cases.

Theia AI: Tools for running tasks and search the workspace

Theia 1.61 introduces two new default tool functions that expand the capabilities of AI agents in Theia-based applications.

The Search Workspace tool allows agents to retrieve relevant content from files across the entire workspace. This capability is especially useful in large codebases where relevant context may be distributed across multiple locations. By enabling agents to search through files directly, this function significantly improves the precision, completeness and speed of code-related suggestions and modifications.

The Run Task tool equips agents with the ability to invoke predefined tasks such as building the project, running tests, or launching an application. This makes it possible for agents to not only propose code changes but also verify them by executing tasks and evaluating the results. It also supports more autonomous behavior by enabling agents to react to runtime outcomes, such as fixing errors discovered during test execution.

Both functions are available to all agents, including custom implementations. Additionally, the AI-powered Theia IDE now includes a new “next” prompt variant for the Theia Coder agent that integrates both of these functions. Users can activate this variant via the AI Configuration view (see screenshot below). In the provided example, Theia Coder is asked to generate a test case and then execute it. During execution, the agent detects a runtime error in the generated code and autonomously corrects it.

Disable installed Plugins and VS Code Extensions

Theia 1.61 supports disabling both installed and built-in plugins directly from the user interface. Previously, users were limited to uninstalling only installed extensions, while built-in plugins remained always active. Users can navigate to any plugin or VS Code extension within the extensions view and choose to disable it.

Disabling a plugin immediately deactivates its functionality after a reload of the front-end. As shown in the screenshot below, for example, disabling the built-in Java extension removes features such as auto-completion for Java files. This can be particularly useful for optimizing performance, avoiding unwanted behaviors, or tailoring the Theia IDE experience to a specific workflow or project need.

New Tree View Example in Yeoman Generator

A new example extension has been added to the Theia Yeoman generator, expanding its capabilities for developers who want to get started with building custom Theia-based applications and extensions. The generator, available here, now includes an example for implementing a custom Theia TreeWidget.

This new example provides a practical demonstration of how to build a TreeWidget using Theia’s tree framework. It highlights several core features such as:

• Lazy loading of tree nodes, enabling efficient handling of large data sets.
• Custom styling and overlays, including the use of colors to visually differentiate elements.
• A structured and extensible layout that can serve as a template for custom navigators, project views, or data explorers.

The following screenshot shows the tree widget example in action.

For a detailed guide on how to compose your own Theia application using the generator and other tools, refer to the official documentation.

As always, the 1.61 release contains much more than described in this article, e.g. for a complete overview of all Theia AI updates in this release, please refer to the collecting epic for 1.61. All these features and improvements (in total 52) were the result of one month of intensive development. Eclipse Theia follows a monthly release schedule. We are looking forward to the next release due next month, stay tuned! To be notified about future releases, follow us on LinkedIn or follow Theia on Twitter and subscribe to our mailing list.

If you are interested in building custom tools or IDEs based on Eclipse Theia, EclipseSource provides consulting and implementation services for Eclipse Theia, for AI-powered tools, as well as for web-based tools in general.

Furthermore, if you want to extend Theia, Theia AI or the Theia IDE with new features or strategically invest into the project, EclipseSource provides sponsored development for Theia, too. Finally, we provide consulting and support for hosting web-based tools in the cloud.

👉 Get in contact with us, to discuss your use case!

👉 Follow us on LinkedIn!