The OpenClaw and Atlassian MCP Server apps are now featured in SoloHost on Pi Desktop, expanding the range of useful self-hosted software that users can more easily install and run on their own computers.
Released alongside the Node 0.6.2 version, the apps show how SoloHost can support different needs, from locally run AI agents to more specialized AI-integrated workflows.
SoloHost enables users to discover, assess, install, and run self-hosted apps through Pi Desktop without having to manually configure as much of the underlying server, Docker, or other technical infrastructure. OpenClaw is a local AI agent that can run on either a local model or external models like ChatGPT and Claude, while the Atlassian MCP Server app provides a local MCP server for connecting compatible AI tools with Jira that helps organize workflows.
OpenClaw on SoloHost: Simpler and More Contained Local AI
OpenClaw is a local AI agent that can run on a user’s own computer and assist with tasks. On computers that meet the hardware requirement to run the agent on a local AI model, OpenClaw can give users greater control over how it interacts with their data and computing environment, while reducing reliance on remote infrastructure for supported tasks. If the agent instead uses a cloud-hosted model, such as Claude or ChatGPT, requests sent to that model are still processed on the model provider’s infrastructure. In either setup, OpenClaw’s memory is created and stored locally on the user’s computer.
When OpenClaw runs through SoloHost, it operates inside a container designed to restrict its default access to the broader computer. Compared with a direct installation outside SoloHost that may provide default wider machine-level access, such as the local file system, this creates a more contained setup with less exposure to unrelated local files and system resources, regardless of whether the configured model is local or cloud.
This containment does not remove every risk or override permissions that a user explicitly grants. If a user authorizes OpenClaw to access an external service or resource, the agent can still act within the scope of that authorization. The benefit of the SoloHost setup is more specific: it limits unnecessary default access to the rest of the local computer while allowing users to grant the explicit permissions needed for particular tasks.
SoloHost also streamlines OpenClaw setup by packaging the application and its required environment for use automatically through Pi Desktop, reducing the amount of manual configuration users would otherwise need to handle.
Adding OpenClaw alongside the already released Hermes local AI agent broadens the range of local AI agent options available through Pi Desktop for users. OpenClaw emphasizes a personal assistant connected across tools and messaging channels, while Hermes places greater emphasis on persistent memory and learning skills from experience.
Atlassian MCP Server on SoloHost: Connecting AI Workflows With Jira
The Atlassian MCP Server on SoloHost addresses a more specialized need for professionals such as developers, product teams, and others who want to integrate AI tools into their Jira workflows.
For an AI tool to interact with a service such as Jira, the two systems need a way to communicate. The Model Context Protocol, or MCP, provides a framework through which compatible AI tools can connect to external systems such as Jira using an MCP server.
Atlassian (Jira) provides its own hosted MCP option, but teams may encounter quota and reliability limitations when using it for internal AI-integrated Jira workflows. This SoloHost app here solves the need for teams that encounter these constraints by locally allowing users to run their own MCP server that is not limited by quota or reliability issues.
Specifically, the Atlassian MCP Server on SoloHost allows the user’s own computer or machine of their choice to run the server locally. An MCP-compatible AI tool can then connect through that local server to work with Jira. Because the connection is based on MCP rather than one particular AI platform, it can work with different MCP-compatible AI tools, such as Cursor, Claude Desktop or Claude Code, and Codex, to make workflow organization on Jira more automated and intelligent.
Expanding SoloHost Utility for Pioneers and the Pi Ecosystem
OpenClaw and the Atlassian MCP Server show how SoloHost can support different types of self-hosted software for different user needs, from local AI agents to more specialized developer utilities.
This gives Pioneers more ways to use their computers for local AI and easier access to other self-hosted applications, further extending the utility of Pi Desktop and Nodes beyond blockchain infrastructure. Through SoloHost, people can benefit from AI while keeping more of their data and computation on their own devices and reducing reliance on third-party remote infrastructure.
At the same time, the two apps demonstrate how developers can create and bring various tools into the Pi ecosystem and make them available to the 420,000+ Pi Node runners and more Desktop users.
As more self-hosted apps are added, SoloHost can continue broadening the role of Pi Desktop as a platform for practical utilities, while giving users more ways to benefit from the computing resources they already operate.
SoloHost remains in beta and uses an open, permissionless publisher flow. Users should assess individual packages and install them at their own risk.