Platform & Extensibility
A workspace you can shape.
Volobox is built as a framework: install capabilities like apps, connect your own tools and models, publish what you build, and extend the workspace itself, in your cloud or ours.
A spectrum, not a ceiling
Configure, compose, or build.
Start with settings. Assemble reusable capabilities without a new codebase. When the need becomes truly yours, extend the same workspace in code.
Library packs
Install agents like apps
Packs bundle agents, skills, commands, and connections. Install first-party packs like the Software Delivery Team (Release Manager, Delivery Coach, Tech Writer), publish your own packs from workspace content, move them between workspaces as files, or pull them from a private registry you host.
Nothing installs automatically; installed items become normal, editable workspace resources; secrets and tokens never travel inside a pack.
Open by design
Connect it to everything
MCP out
Agents call your MCP servers (GitHub and beyond), with calls recorded in the audit trail.
MCP in
External AI tools (Cursor, Claude, and others) read and write Volobox through its own MCP server.
Model freedom
Claude, GPT, Gemini, Kimi, or an open-source model you host, with your own keys, allowlists, and usage visibility.
Secret scopes
Encrypted, scoped secrets for repos, previews, and connections.
Need a workspace app rather than another connection? Build a Custom App visually or by describing it to an agent.
Administration
Admin controls that match the ambition
LLM providers and models, usage and cost visibility, run queues and concurrency, sandbox templates and budgets, MCP allowlisting, working calendar, and an exportable audit log: a detailed admin surface for an AI workspace.
Beyond configuration
When configuration is not enough, we build with you
At the Build end of the spectrum: custom modules, company-specific workflows, and integrations, delivered by the team that builds Volobox.