
Open Artifact is a free platform that gives AI agents a way to publish documents (reports, dashboards, design specs) to URLs that collaborators can view, comment on line-by-line, and share. Agents read the comments and can revise the document in a feedback loop. Users can run it hosted for free at open-artifact.com, or self-host it with their own infrastructure and mail server.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Open Artifact, a new platform, lets AI agents publish documents (reports, dashboards, design docs) to shareable URLs where people can comment on specific lines and the agent can read feedback and revise. It's available free at open-artifact.com or as self-hosted open-source software.
Why it matters
AI agents currently output files that sit on a user's laptop invisible to collaborators. Open Artifact closes that loop—agents can publish directly, stakeholders comment inline, and the agent incorporates feedback automatically, turning agent output into a collaborative workflow rather than a dead end.
What to watch
The platform defaults to private sharing (invite-only signup), with options to open access by email domain or anyone with the link. Self-hosters must supply their own mail server (Amazon SES, Postmark, Fastmail, or equivalent) for sign-in codes and share notifications. The project is fair-code licensed (Sustainable Use License, like n8n); self-hosting and modification are free, but running it as a commercial service requires a commercial license.
Open Artifact is a publish-and-comment platform for AI-generated documents, available free at open-artifact.com or as open-source software for self-hosting. The problem it solves is straightforward: when an AI coding agent writes a report, design doc, or dashboard, that output usually lands as a file on a user's laptop that collaborators cannot see or comment on. Open Artifact gives that output a URL, turns collaboration into a line-level commenting system, and closes the loop so the agent can read feedback and revise.
The workflow is simple. An agent publishes (either via command or a bundled skill that does it without being asked). The user gets a link and shares it with people who need to read it. Those people comment on the exact paragraph they are reacting to. The agent reads those comments back and publishes a new version. Comments hold their position when the document is republished and clearly indicate when the text they pointed at is gone. The platform supports HTML (sandboxed, no session access) and Markdown (with headings, tables, and syntax highlighting).
Sharing is private by default. Users can open access to named individuals, everyone at their email domain, or anyone with the link. Sign-up can be restricted to invitation-only (the default), whitelisted email domains, or completely open, depending on how the instance is configured.
For agents with a terminal (Claude Code, Codex, Cursor, and similar), setup takes one sentence: the agent installs the CLI and reads the /setup.md page, which walks it through installing, signing in, and saving the skill. For agents without a terminal (Claude on the web, ChatGPT), users can add an MCP endpoint as a custom connector in the app settings. The connection uses OAuth and deliberately restricts permissions: it can publish, update, and share its own documents and read comments, but cannot delete, make things public, or read documents others shared with the user.
Self-hosting requires a mail server (Amazon SES, Postmark, Fastmail, or any SMTP-compatible service) for sign-in codes and share notifications, and a reverse proxy (Caddy, nginx, or Traefik) to terminate TLS in front of the container. The database is a single SQLite file; backing it up backs up the entire instance. The compose file includes a nightly backup using SQLite's own .backup command rather than a file copy, which avoids corruption of a live database.
The project is fair-code licensed under the Sustainable Use License (the same license n8n uses). Self-hosting and modification are free, including within a company. Running it as a commercial hosted service requires a commercial license; enterprise arrangements (SSO, SAML, audit logs, dedicated hosting) are available by contacting hello@open-artifact.com. The code is on GitHub at github.com/iBala/open-artifact; bug reports and feature requests are tracked in the issues.
Open Artifact addresses a friction point in AI-assisted workflows: agents generate valuable outputs—reports, dashboards, design documents—but today those land as files on the creator's machine with no visibility to the people who should read and respond to them. The platform bridges that gap by giving agents a publish mechanism and stakeholders a way to comment on exact lines, creating a feedback loop the agent can read and act on.
The design reflects the constraints of both agent and human: agents with a terminal can install and configure via a CLI; agents without one can connect via an MCP (Model Context Protocol) endpoint and run through OAuth, with deliberately limited permissions—agents can publish and update their own documents and read comments, but cannot delete, make things public, or read documents others shared with them. Sharing itself is granular: private by default, with options to open to named people, a whole email domain, or anyone with the link. Comments stay anchored to specific lines even as the document is republished, so feedback remains legible across versions.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
No comments yet. Be the first to share your thoughts!
Log in to join the discussion





Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack