Skip to main content
FlowFn
IntegrationsTemplatesPricingDocsBlogSign inStart free
All posts
Platform-ToolsIntegrationsOAUTHTutorial

Connect Anything — A Guide to FlowFn Platform Tools

Your app rarely lives alone. It needs to post to Slack, charge a card in Stripe, append a row to Google Sheets, send an SMS, open a GitHub issue. Platform Tools is FlowFn's catalog of those connections — over a hundred integrations you authenticate once and then call from any workflow, agent, or MCP server. Connect the outside world once; use it everywhere.

FlowFn Team · Product

16 Jul 2026 · 4 min read

A Platform Tool is a connector to an external service. FlowFn ships a big catalog of them — Slack, Stripe, the Google suite, Notion, Twilio, Shopify, GitHub, databases, and many more — each exposing a set of actions (the operations you can call). You add a connection (a stored, encrypted credential for your team), and from then on any workflow task, agent tool, or MCP server can invoke that tool's actions.

For Bloom & Bean, that means Slack for order alerts and Stripe for payments. Here's the catalog:

The Platform Tools catalog with Slack connected

On the left, browse and search by category — every tool shows how it authenticates (OAuth 2, API key, Access key, Connection string…). On the right, the selected tool's detail: its connection, its capabilities, and its actions. Let's connect one.


Step 1 — Add a connection

A connection is a credential for your team, and the form depends on how the tool authenticates. For an API-key tool like Stripe, you name the connection and paste the key:

The Stripe connect drawer — connection name + API key

For an OAuth tool like Slack, you click Connect, a popup opens to the provider's authorize page, and you approve — no key to copy at all:

The Slack connect drawer — OAuth, with an advanced "use my own OAuth app" toggle

FlowFn provides the OAuth app by default, so connecting Slack is one click; the "Use my own OAuth app (advanced)" toggle is there for teams who want to bring their own client credentials. You can have multiple named connections per tool — a Production and a Staging Stripe, say — and pick which one a given workflow uses.

Once connected, the tool's detail shows the connection as Authorized with its granted permissions (that's Bloom & Bean's Slack in the catalog shot above). Secrets are write-only — after you save, keys and tokens never come back out, only a "connected" state.


Step 2 — Know what a tool can do

Each tool advertises its capabilities and its actions. The capability chips tell you where a tool can be used — as a webhook trigger, from playgrounds, by agents, whether it supports a sandbox or a self-hosted host, or exposes its own MCP server. Below that, the actions: the concrete operations, grouped by category. Click one to see exactly what it takes and returns:

The Post Message action — inputs, outputs, scopes, and agent availability

Slack's Post Message takes a channel and a message (plus an optional thread), returns the message timestamp and channel, needs the chat:write scope, and is available to every agent type. Every action is self-describing like this — which is exactly what lets a workflow map its inputs and an agent's LLM decide when to call it.


Step 3 — Use it

A connected tool is now available in three places:

  • In a workflow — add a Platform tool task, pick the tool, the action, and the connection, then map its inputs from earlier steps. Bloom & Bean's order workflow ends with a Slack Post Message to #kitchen.
  • In an agent — list specific actions in the agent's allowed tools, and the LLM can call them mid-conversation (with a human-in-the-loop approval for the risky ones).
  • In an MCP server — expose selected actions to an external AI client.

In every case, the connection you added is the credential the action runs under, and each platform-tool call costs a small, flat credit fee on top of any work it does.


The escape hatches

Don't see the exact SaaS you need? Three generic tools cover the gaps: Custom API (call any REST endpoint), Custom Webhook (fire an HTTP request at any URL — to receive one, use a workflow webhook trigger instead), and Custom MCP (point at any MCP server). And every tool's detail has a Request a custom action button — the catalog grows from what people ask for.

A note on money: most tools are free to use (you pay only the per-call credit), but some are premium (need a paid plan) or paid (either a one-time purchase or a per-tool subscription). The tool's detail makes that clear before you wire anything up.


Wrap-up

Platform Tools turn "integrate with X" from an SDK-and-OAuth project into: find the tool, add a connection, pick an action. Authenticate once — OAuth in a click, or paste an API key — and that connection is available to every workflow, agent, and MCP server you build, each action self-describing its inputs and outputs. It's the layer that lets your FlowFn app reach out and touch the rest of your stack.

Connect the one tool your app leans on most — for a lot of teams that's Slack or Stripe — then drop its action into a workflow. Next, we'll let your app's own visitors sign in with end-user auth. Connect the outside world once, and everything you build can use it.

Read next

Use-Case
SAAS
Auth

Build a Client Portal with FlowFn

Agencies, consultancies, and B2B teams all hit the same wall: clients want a single place to check their projects, invoices, and files — and emailing PDFs back and forth doesn't scale. A client portal fixes it, but building multi-tenant auth (each client sees only their own data) usually means real engineering. In FlowFn it doesn't: end-user sign-in, per-client data scoping, and a polished portal page are features you assemble, not code you write.

FlowFn Team · 16 Jul 2026 · 3 min read

Use-Case
Devops
Automation

Build an Incident Alerting & On-Call System with FlowFn

When something breaks at 3am, the difference between a blip and an outage is how fast the right person's phone buzzes. You don't need a pricey incident-management platform to get there. FlowFn can catch an alert from any monitor, log the incident, page whoever's on call, escalate if they don't answer, and show the whole team a live status board — built from a webhook, a workflow, a Data Sheet, and a dashboard.

FlowFn Team · 16 Jul 2026 · 3 min read

Use-Case
Internal-Tools
Automation

Build an Internal Ops Tool with FlowFn

Every company runs on a pile of small approvals — refunds, access requests, expenses, discounts — and most of them live in a shared spreadsheet plus a Slack thread nobody can audit. An internal ops tool fixes that: one private table your team works from, action buttons that approve or reject in a click, automatic notifications, and a queue board so nothing rots. FlowFn lets you build it in an afternoon — no internal-tools platform, no login wall to wire up.

FlowFn Team · 16 Jul 2026 · 3 min read

Build your first workflow today

Start free — connect your apps, add AI, and ship an automation you can actually maintain. No credit card required.

Get started free