Reports, documentation, code and analysis your AI writes end up in scratch files and closed chats. Artifacts keeps them in one searchable, versioned place, and your AI can create and read them itself.
The review or the report your AI wrote an hour ago is gone the moment the session closes.
What survives is a file on one laptop, invisible to your next session and to your team.
The next revision overwrites the last one, and the version you liked is not coming back.
Through MCP, your AI tools create, search, retrieve and update artifacts straight from the conversation. Group them by project, type and status, and add metadata you can filter on.
vibexp_io_create_artifact( title: "Q3 incident review", content: "## Timeline ..." )
Every save snapshots the previous content as a numbered version. Browse the history, open any snapshot, diff it against the current content and roll back.
# v5 -> v6 - Report issues found. + Report as: severity, file:line, fix.
Any substantial AI-generated content: code, documentation, work reports, meeting summaries, analysis. Artifacts are Markdown, rendered with syntax highlighting.
In the docs ↗Yes. Over MCP, tools such as Claude Code, Cursor and VS Code can create, search, retrieve and update artifacts directly from a conversation.
In the docs ↗Each artifact keeps a bounded number of versions, 20 by default. The operator of your instance can raise the limit or turn pruning off.
In the docs ↗Stars are how the next person tired of re-explaining their project to an AI finds this. It is the whole of the distribution: no ad budget, no sales team, no hosted plan to upsell.
Free, open source and self-hosted. One command to run it, one to connect your AI tools, and everything else in the docs.