Secure connections. Provision once. Use everywhere.
75+ live enterprise systems, queryable like a database. No copies, no pipelines.
Every system. A database.
One tool, six verbs. The gateway fans out to 75+ systems.
Every system behaves like a database. Workflows become data analysis across systems · which LLMs are good at reasoning over.
The workflow method · describe the schema, query the records, join across systems. Salesforce × Snowflake × Jira in one analysis.
Access controls apply at the data source · and with all access through one CLI, one surface to audit from.
# list → your databases. connection list # describe → schema. Every system has one. connection describe salesforce \ --table Opportunity # query → records, SELECT-shaped. connection query salesforce \ --filter "StageName = 'Closed Lost'" # browse · upload · download → the file-shaped side. connection browse s3 --remote-path reports/2026/ connection download s3 \ --remote-path reports/2026/q2-pipeline.csv # One tool in the agent's context. # 75+ systems · every one of them a database now.
Less surface for the agent. Less data in context. One thing to audit.
150,000 → 2,000 tokens in a single MCP workflow when tools collapse to one interface · Anthropic Engineering · Nov 2025
One place. Every scope.
Access is provisioned for users, agents, and agents working on behalf of users.
Users, agents, and agents working on behalf of users.
Local development and production work the same way.
Every access logged · revoking takes effect immediately.
Provisioned once. Same capability, multiple frameworks.
Pick the surface per use · no re-integration as standards change.
For assistants and MCP clients
Deploying assistants · Claude, ChatGPT, Cursor, Copilot · or building on an MCP client framework.
For your agent sandbox
Use connections directly in the sandbox · composability via shell.
For direct API use
Call the same connections from your own code.
Every system your team actually uses.
Cloud warehouses, databases, SaaS tools, and storage. All accessible through one governed interface.
Don't see your system? We're adding connectors weekly. Request one.
Common questions about Secure connections.
How does Secure connections differ from native MCP servers?
Native MCPs expose each system's own API shape · one server, tool definition, credential, and security review per system. Secure connections treats every system like a database instead: schema you can describe, records you can query, files you can browse · the same verbs everywhere (list, query, describe, browse, upload, download). Provisioned once, reviewed once · not fifty times. Schema is seeded when the connection is provisioned, not rediscovered on every query.
How is a connection scoped?
At provision time. A connection can be owned by one user, shared with named teammates, or shared company-wide · and an agent acts under the scope of the identity it works for. The intent of the task is captured when the connection is provisioned, so access maps to purpose · not to whatever a token happens to permit.
Which surfaces can use a provisioned connection?
All of them. MCP-compatible assistants · Claude, ChatGPT, Cursor, Copilot · through one MCP server; agents that work in code through the connection CLI; the apps and agents you ship through the SDK and API. Same connection, same scope, same audit trail on every path.
What's the authentication model?
Credentials live inside the workspace, encrypted at the boundary, scoped per identity. The LLM never sees a token or API key · Secure connections handles auth at the workspace layer and returns only the data the caller's scope permits. SSO + SCIM cascades govern who can provision connections and who can use them.
Do you support OAuth-only systems?
Yes. OAuth flows are handled per user inside the workspace. Refresh tokens stay encrypted; access tokens never enter the model context. Revocation propagates instantly across every AI surface the user has access to.
How are credentials stored?
In your workspace, encrypted with your KMS keys (Enterprise tier) or our managed KMS (Team tier). For VPC deployments, credentials never leave your cloud · we operate the software, you own the data plane.
Can I build a custom connector?
Yes. Connections is extensible · you can author connectors against any system with an API or database driver. For Enterprise customers we also build new connectors on request as part of the engagement.
How often are new connectors added?
The catalog is past 75 connectors and grows weekly. New connectors are prioritized by customer request frequency. Enterprise customers get custom connectors built into the engagement.
Can I run connectors against my own database?
Yes. Connections supports private databases reachable from inside your VPC · Postgres, MySQL, Mongo, SQL Server, and others · over SSH, private link, or direct in-VPC connection. For air-gap deployments, the entire workspace runs on your hardware.
Same verbs. Every system. Every surface.
Provision once. Use everywhere.
Start free. Provision your first three systems in minutes · and use them from every surface you ship or deploy.
