FastGPT Releases Updated Guidance for Agent Sandbox Runtime
News related to:FastGPT · 2 min read
FastGPT, an open-source AI application platform, has released updated guidance for users of its Agent Sandbox runtime. The guidance covers changes introduced from version 4.16.0 onward and is aimed at ensuring smooth and secure deployments for those using the sandbox feature.
According to the latest release notes, administrators must update the Agent-sandbox-proxy service to version 0.3.0. This update includes optional configuration for PORT and PREVIEW_PORT to split WebSocket and HTTP traffic, which is particularly useful for environments with limited port support. A health check endpoint, `{{host}}/health`, has been added to validate the proxy's connectivity.
The FastGPT platform has also updated several image tags to reflect the new version. The fastgpt-app and fastgpt-pro services have been updated to version 4.16.0, while the fastgpt-plugin has been updated to version 1.1.0. The agent-sandbox-volume and agent-sandbox-proxy components are now at version 0.3.0.
Administrators are advised to adjust environment variables for the fastgpt-app and fastgpt-pro services. New variables include `AGENT_SANDBOX_PREVIEW_PROXY_URL`, which must be configured with the OpenSandbox volume-name prefix. The prefix is listed as `VM_VOLUME_NAME_PREFIX` in the v4.16.0 upgrade note and as `AGENT_SANDBOX_OPENSANDBOX_VOLUME_NAME_PREFIX` in the current environment-variable documentation. Users are encouraged to confirm the correct name against the documentation for the version being deployed.
Several deprecated variables have been removed, including `AGENT_SANDBOX_DISK_MB` and all E2B-related configurations. Users who were previously using the E2B setup must switch to either the opensandbox or sealosdevbox providers and remove the `AGENT_SANDBOX_E2B_API_KEY` variable.
To enhance the flexibility and performance of sandbox deployments, new optional variables have been introduced. These allow administrators to customize the CPU count (default 1), memory limit (default 2048 MiB), storage capacity (default 1 GiB), auto-suspend time (default 60 minutes), and auto-archive time (default 7 days). These settings can be adjusted according to specific deployment requirements.
The press release also notes that mixed version deployments of FastGPT, sandbox proxy, and sandbox components are not supported. While the platform does not enforce cross-origin isolation for sandbox proxy deployments, cross-origin configuration is strongly recommended to mitigate security risks. The release highlights that unsupported use cases include legacy E2B setups, non-standard container orchestration topologies, and custom sandbox provider integrations outside officially supported options.
As of September 7, 2026, the FastGPT GitHub repository, labring/FastGPT, has garnered 29,592 stars and 7,301 forks across 275 releases. The latest version, v4.16.2, was published on September 3, 2026. The platform offers a range of features, including RAG knowledge bases, visual workflows, agent orchestration, Skill, MCP, and multi-channel publishing, available as a cloud service, community self-hosted, or commercial private deployment.