Eclipse Enclave Gives Coding Agents a More Controlled Place to Work
The Eclipse Foundation has released Enclave 1.0, an open-source sandbox designed to let AI coding agents execute tasks in isolated environments with defined access to files, networks and development resources.

The Eclipse Foundation has released version 1.0 of Eclipse Enclave, an open-source runtime and sandbox intended to give autonomous coding agents more freedom to carry out development tasks without granting them unrestricted access to a developer’s machine. Announced on November 19, the project is designed for a growing class of tools that can write and modify code, run terminal commands, install dependencies and execute tests with limited human intervention.
A controlled environment for autonomous work
Enclave runs agent sessions inside isolated containers and allows access to files, networks and development resources to be defined by policy. The aim is to separate an agent’s working environment from sensitive parts of a developer’s system. This matters as coding agents move beyond suggesting snippets and begin performing multi-step tasks that can affect repositories, credentials, software dependencies and connected services.
The Eclipse Foundation says the first release supports several widely used coding agents, including Claude Code, Codex and OpenCode. Its architecture is intended to remain extensible, so additional agents can be integrated rather than tying the sandbox to one model provider or development tool. The project is available as open-source software under the MIT licence.
Why isolation is becoming important
Autonomous agents can be useful when they are allowed to work through a task, test a change and continue iterating. But the same capabilities create operational questions. An agent with broad access may read files it does not need, expose credentials through an external connection or make changes outside the intended project. Organisations also need a way to apply consistent rules when developers use different agents.
Enclave is intended to address part of that problem by placing boundaries around each session. The foundation describes support for parallel agent work, controlled access to external services and mechanisms for protecting API keys and OAuth tokens. Persistent sessions are also part of the design, allowing work to resume without losing its configuration.
What the release does not settle
A sandbox is a control layer, not a guarantee that every agent action is correct or safe. Its value depends on how teams configure permissions, review outputs and manage the software that runs inside the environment. Organisations will still need secure credential handling, code review, dependency checks and clear rules for what an agent may change or send outside a development system.
The launch also reflects a wider shift in software development. As agent tools become more capable, the challenge is no longer only how well they generate code. Developers and security teams must decide how much autonomy to grant, how to observe activity and how to recover from mistakes. A vendor-neutral project could make it easier to apply a common approach across different tools, although adoption will depend on integration quality and the needs of each development team.
A shared foundation for developer choice
By placing Enclave within an open-source foundation, its contributors are inviting developers and organisations to test the runtime, inspect its controls and help shape future integrations. The release does not remove the need for human oversight. Instead, it offers a way to make delegated coding work more bounded, so teams can experiment with autonomous agents while retaining clearer control over the environments in which they operate.
Sources
Related stories

Samsung Recasts Chip Leadership as AI Demand Reshapes Memory Business
Samsung has promoted memory-development executive Hwang Sang-joon to lead its memory business as demand from AI data centres drives a new phase of competition in advanced chips.

US Tightens AI Chip Restrictions, Raising Concerns for Alibaba and Tencent
New US export rules target remote access to advanced American computing chips through overseas data centres, closing a route Chinese technology firms used to obtain AI computing power. Alibaba says the measures could affect some global data-centre expansion plans, while Tencent warns that tighter access may slow future model development.

Sarvam AI Faces Sovereignty Debate Over Chinese Models on Token Factory
Indian AI startup Sarvam is facing questions about its sovereign-AI positioning after adding Chinese-developed models to its Token Factory inference platform. The company says the service runs on privately procured GPUs in India and is intended to give local developers access to a wider choice of models.

Google Brings Gemini Computer Use to Mac, Letting Its AI Handle Multi-Step Desktop Tasks
Google has added computer-use capabilities to its Gemini app for Mac, allowing the assistant to navigate selected desktop apps, local files and websites in a separate window. The English-language rollout is initially limited to the United States.
