Portable context, on one link

The link that speaks for you.

Dump the project in Claude. Send the link. Their agent picks it up — or a person reads the briefing.

Let my agent handle it

Visitors need no account Answers come only from what you approved Unanswered questions return to you

Here’s the problem.

You explain it. Then you explain it again.

Context lives in calls, docs, inboxes, and whoever happened to be in the room. Every handoff loses the nuance.

You explain it once A call, a doc, a voice memo

Someone needs more The nuance never made the handoff

Your stand-in answers The loop ends at the link

How it works

Brain dump in. Briefing out.

Stand.In turns what you know into a reviewed link that keeps working after you send it.

Brief it

Talk, type, paste, attach files, or let your agent dump the project. It listens first.
Approve every fact

Each extracted piece of context waits for your yes. Edit it, keep it private, or cut it.
Send one link

Readers get the briefing first, then a conversation that knows its sources.
Close the gaps

Whatever it could not answer comes back to you as a short list, not a transcript.

The loop

One job: a reviewed handoff.

Dump the project. Approve what may travel. Send one link. People read it; agents ask it. Nothing else is the product.

1 · File

Dump the project.

Ask Claude (or the CLI) to create a Stand In from the repo, notes, Slack, a PDF — whatever you already have. Or drop a doc here.

create_stand_in · npx usestandin create

2 · Review

Approve what may travel.

Mirror shows proposed items. You keep, edit, or reject. Recipients see nothing until you publish.

Mirror · publish

3 · Share

Send one link.

A person reads the briefing and asks grounded questions. Their agent binds the same link. Unknowns come back to you as gaps.

ask_capsule · npx usestandin ask

For agents

Humans get the page. Agents get the context.

Your Claude files the project. The next agent picks up the link.

Two destinations. Your other agent (Cursor, another Claude, the CLI) searches the sources the moment they are filed. Someone else’s agent waits until you publish — then the share link is the credential.

create_stand_in

Claude dumps the repo. Same move on the CLI: npx usestandin create --from .

search_context / read_source

Your next agent, same token. Finds the decision in the sources, not a summary.

ask_capsule

Their agent, the published link. No account. CLI: npx usestandin ask <link> “…”

Hand it off from Claude

MCP and the CLI are the same path. File it, send the link, they pick it up.

1. File it
Connect the authoring server in Claude, then say “create a Stand In from this repo.” Or run npx usestandin create --from .
2. Send the link — or keep it
Your Cursor can search_context immediately. A collaborator’s agent sees nothing until you approve and publish.
3. They pick it up
They add the respond server (no key) and paste the link. Or: npx usestandin ask <link> "why session cookies"

Your Claude · files it

Add the authoring server. Claude signs in in the browser on first use — no token to paste:

claude mcp add --transport http standin \
  https://standin.app/api/mcp

Or open Claude connectors. CLI: npx usestandin create --from . after Connect on home.

Their agent · picks it up

No account, no key. The published link is the credential:

claude mcp add --transport http standin-respond \
  https://standin.app/api/respond/mcp

Then “read this: https://standin.app/…” or npx usestandin ask <link> "..."

Step-by-step for every client: /docs/connect. Cursor JSON at /install/cursor. Protocol: /llms.txt.

Pricing

Start free. Make it yours when you're ready.

Everything is free while we are in beta. Here are the plans as they will land.

Free

Your first stand-in.

One published stand-in
Full capture, briefing, and ask
Your asks inbox
About 100 answered questions a month
Carries the Stand In mark

Start free

Stop repeating yourself.

Brief it once. Send the link. Let it do the explaining.

Create a stand.in · 2 min