One governed endpoint for the AI tools your team uses.
Combiner is a managed MCP service that turns useful tools into approved, metered, workspace-ready capabilities. Connect a supported MCP client to one URL. Owners decide who can use what, the workspace shares one credit budget, and every call is recorded.
https://combiner.to/mcp
Your team is already connecting AI tools. Nobody is controlling them.
Your team points AI clients at tools every day. There is no approved tool set, no shared spend limit, no per-person access, and no record of what ran or what was blocked. The tools are useful. The lack of control is the problem.
Anyone can wire up anything. There is no agreed list of tools the team is allowed to use.
Paid calls run with no shared limit and no single place to watch spend.
Everyone gets everything, or access is managed by hand, tool by tool.
There is no record of which call ran, who made it, or why one was blocked.
One managed service in front of your tools.
Your team connects an AI client to a single endpoint. Combiner authenticates the person, finds their workspace, shows only the tools they are allowed to use, meters paid calls against a shared budget, and records every call. The workspace decides access, not the URL.
MCP is the open standard that lets AI clients call external tools. Combiner is the layer that decides which tools your team can call, what they cost, and who is allowed.
- 1Authenticate the personOAuth sign-in through WorkOS.
- 2Resolve the workspaceMembership and role decide what is visible.
- 3Show only granted toolsService Bundles scope what each member can call.
- 4Meter and recordPaid calls draw on the shared budget. Every call is audited.
Access comes in bundles you grant, not a firehose you police.
Tools are grouped into Service Bundles. An owner grants a bundle to a member, and that member can use exactly the tools in it. Nothing more. Grants are additive, scoped, and reversible, and every grant and revoke is recorded. Least privilege is the default, not a setting you have to find.
illustrative
A bundle groups a set of tools. Grant it to a member and they can use exactly these, nothing more.
One governed endpoint for any team's tools.
Combiner is not a catalog you dig through. We curate and run a focused set of governed tools, and we shape access around the work a team actually does. The model is the same whatever your team works on: approved tools, a shared budget, and a record of every call. We add services deliberately, so every tool stays governed, metered, and auditable.
Access maps to what a team does, from research and operations to client work, not a generic tool dump.
Every tool is picked, priced, run, and tested before it ships, so each one stays governed and auditable.
The same approval, budget, and audit model carries across teams and the work they do.
Tell us what your team needs during early access, and we will tell you what is available today and what we can shape next.
One shared budget. Real prices. No surprise bills.
Your workspace shares one credit pool. Every paid tool has a clear credit price, and free tools stay free. The owner sees usage and denials as they happen and can cap or revoke access at any time. If a tool call fails upstream, the credits are refunded, not charged.
Built so you can see and prove what happened.
Sign-in uses OAuth through WorkOS. Every call your team makes, allowed or denied, is recorded with the reason, the tool, the member, and a request id. Logging is metadata only; Combiner does not store the content of what was fetched. Workspace credentials are encrypted and never shown in plaintext after you save them.
| Time | Member | Tool | Outcome | Reason |
|---|---|---|---|---|
| 09:41 | Member | fetch |
Allowed | charged 1 credit |
| 09:43 | Member | search |
Denied | no_credits |
| 09:44 | Owner | read |
Allowed | free tool |
Connect once. Govern from day one.
- 1Connect your clientCreate a workspace and connect a supported MCP client to the Combiner endpoint.
- 2Grant access and set a budgetAs owner, grant a Service Bundle to each member and set the shared credit budget.
- 3Your team worksMembers use the granted tools through their own client.
- 4Watch and revokeWatch usage and denials, and revoke access any time, with a record of every change.
https://combiner.to/mcp
{
"mcpServers": {
"combiner": {
"url": "https://combiner.to/mcp"
}
}
}
Most platforms sell more tools. Combiner sells control.
Other platforms compete on how many tools they connect and how fast you can wire them in. Governance and spend control sit behind a sales contact form. Combiner inverts that. Access, budget, and audit are the product. Fewer tools, each one approved, metered, and on the record.
Service Bundles are the grant unit. Members get exactly the tools an owner approved, nothing more.
One budget the owner funds and watches, with clear per-tool prices and a free versus paid split.
Every call, allowed or denied, is on the record with the reason, the tool, and the member.
Questions teams ask.
What is MCP, and do I need to understand it?
MCP is the open standard that lets AI clients call external tools. Combiner handles the protocol. You manage access and spend.
Which AI clients work?
Combiner is verified today with Claude Web and Claude Desktop, plus the FastMCP Inspector for testing. ChatGPT is not verified end to end yet, so it is not listed as supported.
What does it cost?
Credits are the internal unit. Paid tools have a clear credit price and free tools stay free. Pricing is handled per team during early access.
What tools can my team use?
Combiner runs a curated, governed set of tools and adds services deliberately, so every tool stays metered and auditable. Tell us what your team needs during early access, and we will tell you what is available today.
Where does my data go, and what do you log?
Logging is metadata only. Combiner does not store the content of what was fetched. Workspace credentials are encrypted and never shown in plaintext after you save them.
Is Combiner only for large teams?
No. A personal workspace is a one-person workspace. Teams add shared credits, grants, and audit on top of the same model.
Bring your team's AI tools under control.
Request access and connect your first client, or see how it works first.