Incredibuild Introduces Islo Sandbox for Secure AI Coding Agents

Incredibuild has launched Islo, a sandbox platform designed to safely run and manage AI coding agents in isolated environments announced in a press release. The system provides enterprise security controls and performance isolation, allowing engineering teams to operate coding agents continuously without exposing sensitive systems.

Islo offers isolated execution environments with granular policy management that define how agents interact with codebases, infrastructure, and external tools such as GitHub Copilot and Jira. Each agent runs in its own dedicated environment, separate from developer machines, ensuring reproducibility and continuous operation.

The platform includes persistent execution contexts, integrated acceleration based on Incredibuild’s existing technology, and observability tools for monitoring agent activities. It supports secure connections to enterprise systems while maintaining control over agent permissions and data access.

In addition to development use cases, Islo integrates with the open source Harbor Framework to support AI research, enabling large scale evaluations, benchmarks, and reinforcement learning rollouts within controlled environments. Developers can access the platform at islo.dev.