Bring in what matters
Start with a document, a web link, or a piece of text. You can also connect supported sources, import past AI conversations, or add a code repository. Choose the material you want to include; you do not need to organize everything first.
One useful project is enough to begin. A brief, a few meeting notes, and a record of recent decisions give OpenContext something concrete to work with. You can add more as the project grows.
Build a shared understanding
OpenContext brings people, projects, decisions, and ideas together in a connected knowledge library. Related material can contribute to the same topic, so the understanding of a project can grow across documents and conversations.
For example, a project brief might explain the goal, meeting notes might record a change in direction, and a later update might explain why. OpenContext connects that context so you can follow the story and return to the supporting sources.
You can browse the library, review what was added, and correct information when needed. Source links help you assess an answer; they do not make every interpretation automatically correct.
Put your context to work
Search for a person, project, or decision. Ask a question in chat and follow its citations. Explore suggested connections, then choose which ones are worth keeping.
Your external AI tools can use the same knowledge through the command-line interface or the Model Context Protocol (MCP), a standard for connecting AI tools to other services. Coding agents can draw on an indexed repository's context, and new decisions can be contributed back for future work.
Instead of rebuilding the background in every conversation, you can begin with what is already known. See the agent guide for installation, adding material, and connecting your tools.
Stay in control
Your knowledge lives in a local vault on the computer running OpenContext. You can inspect it, review its history, undo changes, and export it. The app shows ongoing work and lets you stop a run.
Privacy checks help identify sensitive content, with findings and held items available for review. These checks have limits. When you use AI features, the content needed for that work goes to the model destination you configure, which may be a cloud provider. The privacy page explains the controls and data handling in more detail.
Keep it useful as work changes
Add new material and refresh connected sources as your work moves forward. Review the resulting updates, ask follow-up questions, and save the decisions you want future conversations to remember.
For code repositories, refresh the project context as the code changes and check its freshness before relying on it. OpenContext gives an agent a starting point; the current source files remain the reference for the code itself.
Ready to try it? Follow the installation guide, then bring in one useful document. The frequently asked questions cover common setup and usage questions.