TL;DR: You can send a contract for signature by typing one sentence in Claude Desktop. Connect it once to the BoldSign streamable MCP server, and Claude gains a set of BoldSign actions it can use on your say-so. Claude finds the template, fills in who the document goes to, asks you to confirm, and sends it. Claude does not sign anything and cannot change what the contract says. Your recipient signs it themselves, from their own inbox. Setup takes about 10 minutes, with no Node.js, no terminal commands, and no code.
Claude can talk about contracts. Now let it send them
You have probably already used Claude to help with an agreement: drafting the wording, checking a clause, or deciding who needs to sign. Then, when the document is ready, you open a separate app, upload the file, set up the signers, and send it by hand.
That extra trip is the problem this post solves. By the end of this guide, you will type one sentence in Claude Desktop and a real contract will land in the recipient’s inbox, with nothing installed on your computer and no code written.
Time needed: about 10 minutes for setup, then seconds for every send after that.
If you are a developer who wants to embed this into a product or build a custom agent workflow, read the developer guide instead. This post is for anyone who uses Claude Desktop and wants it to handle the sending step.
Key takeaways
- Connect Claude Desktop to BoldSign in about 10 minutes using OAuth
- Send contracts from chat using a single natural-language prompt
- Use existing BoldSign templates for secure, controlled document delivery
- Track status, send reminders, and cancel documents without leaving Claude
- No coding, local MCP server setup, or terminal commands required
- Recipients sign documents themselves, preserving the standard BoldSign signing experience and audit trail
How it works in plain language
MCP stands for Model Context Protocol. It is an open standard that lets AI assistants use outside tools, in the same way your phone uses apps.
BoldSign runs a hosted, streamable MCP server at a public URL. When you connect Claude Desktop to it, Claude gains a set of BoldSign actions: send a document, check its status, send a reminder, cancel it. You never touch those actions directly. You just describe what you want in chat, and Claude handles the rest.
Because the server is hosted by BoldSign, there is nothing to install or maintain on your computer. The connection is a URL and a one-time sign-in to your BoldSign account, added once in Claude Desktop settings.
One thing to be clear about: Claude does not sign. Talk of AI signing gives the wrong impression of what is happening here. Claude prepares and sends the document. The person you name as the recipient opens a normal BoldSign email and applies their own signature, with the same consent step and the same audit trail as any other electronic signature. What you are removing is the trip to another app, not the human decision at the end.
What you need before you start
A BoldSign account. A free trial or sandbox account is enough to complete this guide. Use the sandbox while you practice so no real contract goes out. Make sure you have at least one template in your account before you start; an NDA with one signer role is ideal for testing.
An OAuth connection in BoldSign. You’ll create this in your BoldSign account settings during Step 1. It provides a Client ID and Client Secret that you’ll paste into Claude Desktop.
Claude Desktop. Download it from claude.ai/download if you have not already. Custom connectors work on Free, Pro, Max, Team, and Enterprise plans, so BoldSign is available regardless of which plan you’re on. One thing to know if you’re on a Free plan: you’re limited to one custom connector in total. If you already have a different connector added, you’ll need to remove it before you can add BoldSign.
The right regional URL. BoldSign runs a separate MCP server URL for each region: US, EU, CA, and AU. Yours has to match the region your BoldSign account is in, or the connection will fail. Check your account’s region first, then use the matching URL in Step 2. Full details, including all four URLs, are in the BoldSign MCP server documentation.
That is the complete list. No Node.js, no terminal, no developer tools.
Connecting Claude Desktop to BoldSign
Steps 1 and 2 below are written for one person connecting their own Claude Desktop, which works as written if you’re on Free, Pro, or Max. If your organization is on a Team or Enterprise plan, individual members can’t add a custom connector themselves: an Owner or Primary Owner has to complete Steps 1 and 2 first, using Organization settings > Connectors instead of Customize > Connectors. That makes BoldSign appear automatically in every team member’s connector list. From there, each person only needs to sign in with their own BoldSign account, no Client ID or Secret required on their end.
Step 1: Create an OAuth connection in BoldSign
If you already created this connection, skip to Step 2.
1. Log into the BoldSign web app
2. Go to API > OAuth Apps, click Create App, and give it a descriptive name, such as Claude Desktop.
3. In the redirect URLs, add both redirect URIs listed on BoldSign’s Claude Web connector guide.
4. Copy the Client ID and Client Secret it generates. The secret is shown only once; if you close the page before copying it, generate a new one.
Step 2: Add BoldSign to Claude Desktop
1. Open Claude Desktop
2. Go to Settings by clicking your name or the gear icon, then select Customize.
3. Select Connectors
4. Click Add, then Add custom connector.
5. Add name and enter the BoldSign MCP server URL for your account region. In this example, I’ll use the US region endpoint: https://mcp.boldsign.com/mcp. If your BoldSign account is hosted in another region, use the corresponding regional endpoint listed in the BoldSign MCP Server Documentation.
6. Click Continue. In the OAuth Client section, enter the Client ID and Client Secret created in Step 1, and then click Add.
7. Click Connect. Your browser opens a BoldSign sign-in screen; sign in and approve access for Claude Desktop
8. Save the connector and fully quit and reopen Claude Desktop
After Claude Desktop restarts, open a new conversation. You should see BoldSign listed in the available tools. If you do not, try the troubleshooting section at the end of this guide.
The streamable server runs on BoldSign’s infrastructure, not on your computer. There is no process to start, no config file to edit, and no terminal command to run.
Step 3: Prepare a contract template in BoldSign
Claude can send documents only from existing BoldSign templates. It cannot change contract wording, upload new files, or edit fields. This is intentional and keeps legal content under your control.
If you already have a template, move to Step 4. If not, here is how to create a simple one:
- In the BoldSign web app, click Create New → Create New Template
- Upload a PDF such as a simple NDA and add one signer role, and call it Recipient or Signer
- Place form fields for that role and save the template
The template name matters. Claude will find templates by the name you give them. Use something clear and memorable, like Mutual NDA or Contractor Agreement, because that is how you will refer to it in chat.
Send your first contract from chat
Open a conversation in Claude Desktop and type:
Send our NDA template to Sarah Wilson at [email protected] Claude will find the template, confirm the recipient details with you, and wait for you to say yes before sending. You should see something like:
I found the template “Mutual NDA” in your BoldSign account.
I’ll send it to Sarah Wilson at [email protected].
Shall I go ahead?
You: yes
Sent. The Mutual NDA is on its way to [email protected] for signature.
Document ID: 10793399-xxxx-xxxx-xxxx-eaa3dfa2123
Recipient: Sarah ([email protected])
Use a test email address you control as the recipient so you can see the signature request arrive. Once you type yes and Claude confirms the send, check that inbox. The BoldSign email will be there, looking exactly as if a person had sent it from the BoldSign web app.
Track, remind, and cancel from the same chat
Sending is just the start. Try these follow-up prompts in the same conversation:
Check status
What is the status of the NDA I just sent to Sarah?Claude calls the document detail tool and reports back in plain language: whether the document has been viewed, signed, or is still waiting.
Send a reminder
Send Sarah a reminder about the NDA.Claude sends the reminder through BoldSign. The recipient receives a standard follow-up email, the same one you would get if you clicked Remind from the BoldSign web app.
Cancel a document
Cancel the NDA I sent to Sarah.Claude confirms the cancellation with you before acting, then calls the revoke tool. The document is cancelled and Sarah’s signing link stops working.
All of these work from the same chat window. You do not need to open BoldSign, switch apps, or remember document IDs.
What Claude can do once it is connected
Fourteen BoldSign actions become available to Claude. You will use six of them almost all of the time:
- Send a document from a template to someone you name
- List your templates, so you can ask which ones you have
- Check a document’s status, including who has and has not signed
- Send a reminder to a recipient who has not signed yet
- Cancel a document that has already gone out
- List your documents with where each one stands
The rest cover contacts, users, and teams. Claude picks up all of them on its own when it connects, so there is nothing to switch on individually. The full list, with what each one does, is in the BoldSign MCP server documentation.
Everyday use cases for teams
Sales: NDA in the moment. When a deal conversation needs an NDA, send it without switching apps. “Send our NDA template to Alex Chen at [email protected]” takes one sentence and no app switching.
HR: offer letters and onboarding forms. Send an offer letter from chat at the end of an interview debrief. Track whether it has been signed without logging into BoldSign.
Support: authorization forms during a ticket. While working a support ticket, send a warranty or authorization form to the customer from the same conversation. Try the live AI Warranty Claim Agent demo at demos.boldsign.com to see this pattern in action.
Keeping control: safety for AI-sent documents
The most common question is: can Claude change the contract before sending? The answer is no. Claude can only send documents from templates you have already approved in BoldSign. It fills in who receives the document, not what it says.
Claude confirms before it sends. Every send and every cancellation goes through a confirmation step in the chat. Claude tells you the template name and recipient before acting, and waits for you to say yes.
Use a dedicated OAuth connection. An OAuth app created specifically for Claude Desktop can be revoked on its own. If you ever want to disconnect Claude from BoldSign, revoke that connection in your BoldSign account settings and access stops immediately.
Practice in the sandbox. BoldSign sandbox accounts behave exactly like production accounts but do not send real signature requests. Sign in with your sandbox account while you get comfortable, and connect a production account only when you are ready.
Every action is logged. BoldSign records an audit trail for every document sent, including who or what triggered the send. Your BoldSign dashboard shows the full history.
Are AI-sent contracts legally valid
Legal validity depends on the applicable laws in your jurisdiction and the type of document being signed. In general, legal enforceability comes from the eSignature process and the signer’s consent, not from whether the document was sent by a person or initiated through AI.
Documents sent through the BoldSign MCP server are processed the same way as documents sent through the BoldSign web app, with the same signing experience, audit trail, and completion certificate.
If you have specific legal, regulatory, or compliance requirements, consult qualified legal counsel. For information about the standards and regulations supported by BoldSign, refer to BoldSign’s compliance documentation.
Troubleshooting
If something is not working, here are the most common causes:
| Problem | Fix |
| BoldSign not in tools list | Fully quit and reopen Claude Desktop after adding the connector, not just close the window. On Mac, use Cmd+Q; on Windows, right-click the taskbar icon and select Quit. |
| Connector fails to connect, or shows an authorization error | The Client ID or Client Secret may have been copied incorrectly, or the redirect URI registered in BoldSign may not match the one provided by Claude Desktop. Regenerate the credentials in BoldSign, confirm the redirect URI matches exactly, and reconnect. |
| OAuth sign-in screen doesn’t appear, or the connection gets stuck | Check that pop-ups or redirects aren’t blocked by your browser, and that you’re logging into the correct BoldSign account (sandbox vs. production) before approving access. |
| Template not found | Refer to the template by the exact name it has in BoldSign. Sandbox and production accounts have separate templates, so make sure you are using the right account for the right templates. |
| Claude sends without confirming | This should not happen by default. If it does, start a new conversation and try again, because long conversations can occasionally drift from the expected behavior. |
| Connector settings are not visible | Custom connectors work on every Claude plan, so this usually means you’re on a Free plan and already at the one-connector limit. Remove your other custom connector, or upgrade, then add BoldSign. |
| Connector shows as connected, but every BoldSign tool call fails or returns an authorization error. | Confirm the MCP server URL matches your BoldSign account’s region (see What you need before you start). A US URL against an EU, CA, or AU account will fail even with correct Client ID and Secret. Reconnect using the matching regional URL. |
Quick-start checklist
Before you close this guide, run through this list:
- BoldSign account created (sandbox recommended for first run)
- At least one template in your BoldSign account with a signer role
- OAuth connection created in BoldSign, with Client ID and Client Secret copied
- BoldSign connector added in Claude Desktop settings with the correct regional URL from Step 2, with the Client ID and Secret entered under Advanced settings
- Signed in to BoldSign and approved access when prompted
- Claude Desktop fully quit and reopened
- BoldSign tools visible in a new conversation
- First test send completed to an email address you control
Once you can check all of those, you are set. Every contract send from that point is one sentence in chat.
You just turned Claude into a contract-sending assistant, with nothing installed
You connected Claude Desktop to the BoldSign streamable MCP server with a one-time BoldSign sign-in, sent a real contract by typing one sentence, and managed it through status, reminders, and cancellation from the same chat. The setup took about 10 minutes and will not need to be touched again.
That is the practical point of the BoldSign streamable MCP server: the connection is remote, sign-in happens once through OAuth, and the workflow benefit starts immediately.
For a more advanced example of MCP-powered eSignature, try the AI Warranty Claim Agent demo at demos.boldsign.com. It uses the same architecture, applied to a full support scenario where a customer chat ends with a signed warranty form.
Building something larger, such as an agent that runs automatically, or a feature inside your product? The developer guide walks through the Claude API and BoldSign MCP integration with full code, approval gates, and production patterns.
Ready to try it? Sign up for a free BoldSign sandbox account, create your first template, and connect Claude Desktop using the steps above.
FAQ
Does Claude sign the contract?
No. Claude prepares and sends the document. The recipient you specify signs it from their own inbox, just as they would with a request sent through the BoldSign web app. The signature, consent, and audit trail all belong to the recipient.
Do I need any coding skills to set this up?
No. You add a URL, paste in a Client ID and Client Secret, and sign in through a secure BoldSign screen. That is the entire setup. There is no code to write.
What is the difference between this and the local MCP server?
The local server runs on your computer and requires Node.js, a config file edit, and a terminal command. The streamable server runs on BoldSign’s infrastructure and is accessed through a regional MCP endpoint. This guide uses the streamable MCP server rather than the local MCP server.
Can Claude change what the contract says before sending?
No. Claude can only send from templates you have already created in BoldSign. It supplies the recipient’s name and email but cannot edit the document content.
Which Claude plans support this?
Custom connectors work on Free, Pro, Max, Team, and Enterprise plans, so BoldSign is available on any of them.
Can I use this for my whole team?
On Team and Enterprise plans, yes, and an org admin can set it up once for everyone: see “Setting this up for a team” above. On Free, Pro, or Max, each person connects their own Claude Desktop individually. Either way, each person still signs in with their own BoldSign account. For team-wide automation or embedding the capability into a product, see the developer guide, which covers the Claude API and custom agent builds.
Is this free to use?
The BoldSign MCP server is free to connect. Document sending uses your BoldSign plan’s allowance. Anthropic API credits are not needed for Claude Desktop, because you use your Claude subscription.
What happens if I revoke access?
Claude Desktop loses access to BoldSign immediately. You can revoke the connection at any time from your BoldSign account settings. To reconnect, sign in again from Claude Desktop’s connector settings.
