ReferenceProductConnection types
Connection types
Work out which connection type reaches a particular system, what each asks for, and what to do when a system fits none of them.
Skills for your coding agent
See it running: Trip planner
When you create a connection you pick what you're connecting to. The type decides what credential it asks for and how its operations are found.
The types#
Your coding agent creates every type, so you describe the system and it picks the type. In the Studio, Connections → Copy prompt to create gives you a prompt to start from.
| Type | What it reaches | What it asks for |
|---|---|---|
| HTTP API | Anything with an HTTP API | A token sent as Authorization: Bearer <token>, or for another header, an empty token and up to three header rows (name and value). Your coding agent can also set up HTTP Basic auth, or none for a public API |
| MCP server | The tools a remote MCP server exposes | A token or headers, or Connect in your browser if the server uses OAuth |
| AI model | One model, under limits an admin sets | Nothing. Mindset uses its own access to the model |
| Postgres database | A database, through queries you enable one at a time | A connection string |
| Google Sheet | One spreadsheet: read a range, append a row | A Google Cloud service account key |
| Knowledge base | A search server that holds your documents | Whatever an MCP server would: a token, headers or a browser sign-in. The server must speak MCP and expose exactly one tool, named rag_search |
To connect any of them from your coding agent, see Connect a system.
What each one is for#
HTTP API is where most connections end up. The finance system in the invoice workflow is one. Its operations are set up one per method and path.
MCP server reaches things Mindset doesn't do itself: reading text from a scanned invoice, producing a spreadsheet, or a product whose vendor ships an MCP server. Each tool becomes an operation, and from then on it behaves like any other operation.
AI model is how functions and scripts call a model. An admin picks one model and sets a Size cap (maximum output tokens per call) and PII protection, which is on by default. A function step or a script phase names a model connection, not a model, so it can't pick a different model than the one it was given. The PII protection switch only matters while your org's personal data setting is on. See Audit and personal data.
Postgres database is for lookups and joins over more data than a sheet holds: supplier payment terms and two years of invoice history.
Google Sheet is for small tables a person also maintains by hand: who approves supplier queries, or how much variance each supplier is allowed.
Knowledge base grounds judgment in your own material. In the invoice workflow, the Assess phase searches the supplier contracts for the clause covering a price difference. Mindset doesn't load or index documents itself.
Google#
Google is set up by name through your AI tool. You pick the service and whether it's Shared or Individual.
- Google Calendar: list, search, read, create, change and delete events.
- Google Drive: create and read Google Docs and folders, export a Doc as a PDF, rename, share and trash files. It sees only the files it created or that the person opened with it.
- Gmail: sends email from the connected account. It can't read the mailbox. Sending is a write, so it follows your org's approval setting.
Google runs on your own OAuth client. See Connect a system.
The Slack app#
A Slack connection is an HTTP API connection on the Slack app adapter. It lets a script post a decision to Slack with buttons and receive the click back, through an $ask step. Create it from your coding agent, which asks for three values from a Slack app you create in your own workspace: the bot token, the signing secret and the app ID. The connection's Configuration tab then shows Let a person answer in Slack, with the app manifest and the Request URL to copy into Slack.
For the setup step by step, see Post to Slack or send email.
How credentials are handled#
Whatever the type, Mindset holds the credential and adds it to each call, and it never reaches the agent or the model. How that's enforced is on Security and approval.
When the system you need isn't listed#
Ask three questions, in this order:
- Does it have an API? Most systems do, including many whose vendor sells an integration and never mentions the API underneath. If it has one, use HTTP API.
- Is there a remote MCP server for it? Vendors and communities publish them for many products. A server someone on your team runs on their own laptop can't be connected. It needs a remote address.
- Is it only reachable by logging into a website and clicking through its screens? Then Mindset can't reach it. There is no type that drives a browser. Ask the vendor whether there's an API behind those screens. There usually is.
Before an agent can call an operation#
The connection must be live, and the operation enabled, available to agents, approved if it writes, and assigned in the agent's active version. See Connections and operations.
Limits#
- An HTTP API that needs OAuth can't be connected yet, except Google. Use an API key or header auth. MCP servers support OAuth sign-in, Shared or Individual.
- Only remote MCP servers can be connected. A local (stdio) server can't.
- An AI model connection has no cost cap or data residency control that Mindset enforces. The size cap and PII protection are the limits it applies.
- A knowledge base can't be filled from Mindset. Its documents live on the search server.
You're done when#
- You know which type each system in your workflow uses.
- You have a credential for each one, scoped to what its operations need and pointed at the right account.
- For anything that didn't fit, you know whether it has an API or a remote MCP server, or that it has neither.