Claude Code · MCP
Claude Code wrote it.
Now send it to someone.
Claude Code is very good at building a report, a dashboard or a working prototype as a single HTML file. Getting that file in front of the people who asked for it is the part that is still manual.
No card required · Unlimited viewers · Secure by default
The situation
What happens to Claude Code's output today
The artifact is stuck on your machine
It renders beautifully in your terminal's preview and nowhere else. Sending it means attaching a file, and an attachment is a copy the moment it leaves.
So it becomes an attachment, and then a stale one
You email index.html. Someone forwards it. A week later Claude rebuilds it with this month's numbers and there is no way to reach the people holding last month's.
And nothing was ever access-controlled
A file has no audience. Anyone who receives it can open it, forward it, and keep it after they leave — which is why this output rarely gets to leave the engineering team at all.
Setup
Connect Claude Code
Ask Claude to build a report, then “publish this to Finance” — without leaving the terminal.
- 1.
Add the server
claude mcp add --transport http deliverd https://deliverd.dev/api/mcp
Claude Code opens a browser the first time it is used, so you can pick the organisation and approve access.
No token is written anywhere. Deliverd implements the MCP authorization spec, so the client registers itself and you approve it in a browser — and removing someone from your organisation cuts their connection off at once, rather than whenever a token happens to expire.
What you get
Once it is connected
Everything below is what happens to a report after Claude Code publishes it.
A URL instead of a file
The report lands on an address you can send. It stays put, so a link you shared last month still opens the report — not a stale copy of it.
Behind your own sign-in
Readers authenticate with the identity your organisation already uses, including SAML single sign-on. Nobody gets a password to remember or a file to lose.
You choose who, in plain English
“Finance team”, “Sarah Jones”, “everyone”. Ambiguous phrases come back as a list to pick from rather than a guess, so a board pack cannot land in the wrong inbox.
Every version kept
Publishing again updates the same URL and keeps what was there before. You can see what changed, and roll back if the new one is wrong.
Comments on the passage, not the page
A reader selects a sentence and comments on it. Your agent can read those threads, fix what they point at, and mark them resolved.
A record of who opened it
Views, versions, access changes and share links are all written to an audit trail you can export.
Questions
Claude Code and Deliverd.
Do I need an API token?
No. The server implements the MCP authorization spec, so Claude Code registers itself and opens a browser for you to approve the connection. Nothing writes a credential to disk.
Does this work with Claude Desktop and the web app too?
Yes — add the same address under Settings → Connectors. It is the same server and the same tools.
What can Claude actually do once it is connected?
Publish a report, update one in place, rename it, file it into a workspace, search what exists, read reader comments and resolve them, and check which recurring reports are waiting on it.
Can it publish something publicly by accident?
An agent identity can never grant public access — that is refused at the service, not just hidden in the UI. Everything else runs through your organisation's publishing policy.
Stop sending Claude Code’s output around as files.
Publish it to an address instead, and let the people who need it open it with the sign-in they already have.