Central API management

The integration infrastructure your team and agents run on.

Every API your product depends on — first-party and third-party — behind one access point. Your services reach it over HTTP; your agents reach the same thing over MCP. Compose them when one call isn't enough, see what's healthy in a single view, and change what's underneath without shipping code.

Or browse the apps and read the specs first.

Your teamStudio, one login
Your APIone bearer key
Your productper-tenant, federated
Your agentstools, not credentials
w6wrouting Your team

Endpoints

one door, one key

Functions & Workflows

compose several calls

Connections & vault

credentials stay here

Health & monitoring

latency, errors, uptime

vendor credentials never cross this line

Facebook Pages
YouTube
WhatsApp
Gmail
Google Drive
WordPress
LinkedIn
Google Calendar
Google Docs
Google Sheets
OpenAIOpenAI
Telegram
Microsoft Outlook
Microsoft Excel
Spotify
Twitter (X)Twitter (X)
Reddit
Microsoft Teams
PayPal
Microsoft OneDriveMicrosoft OneDrive

+ thousands of integrations

What does w6w actually do?

w6w is central API management: the APIs your product runs on — the ones you own and the ones you buy — reached through one access point, with one credential. Four things have to be true for that to hold, and the product is the four together. Buy them separately and you get four tools that don't know about each other, because the unit being managed is an API — not a workflow, and not a credential.

One front door

Many APIs, one access point, one credential.

Your callers reach one Endpoint, holding a key of yours rather than a vendor's. The credential stays on the host, in the vault — nothing you ship carries a token, and rotating one breaks no caller.

Composition

For when one call isn't enough.

A Function is one named operation, whichever vendor sits underneath it. A Workflow is a durable graph with branching and sub-workflow calls. Both answer at the same Endpoint, so a caller never has to know which it got.

Visibility

Your APIs and your vendors', in one view.

Every app declares its own health probes — vendor up, connection live, quota left — so a failing integration announces itself. Every outbound call is recorded, with credential fields redacted by name.

Plug & play

Change what's underneath without shipping code.

Re-point "send email" from SendGrid to Brevo and no caller notices the change. Pin a version per tenant, and configure what a running integration does instead of cutting a release and waiting for it.

Multi-tenant by construction

Your customers connect their accounts inside your product.

Federate each tenant to its own identity provider, entitle it to exactly the apps you allow, and keep every connection scoped to the tenant that made it. Nobody is sent to a page of ours to sign in.

Any API, not just ours

The catalog is a convenience, not a gate.

An app is a published spec — a manifest, actions, auth methods, health checks. Import one from a GitHub URL, pinned by content address, with no vendor in the loop and no queue to join.

Who is this for?

Teams carrying integration sprawl

Thirty vendors, credentials scattered across a vault and application code, and no single place that knows whether any of it works. Put them behind one door and change them without a release.

Platforms that integrate on behalf of their customers

Multi-tenant from the ground up: federate each tenant's identity, lock every tenant to the apps you allow, and let customers connect their accounts inside your product — never ours.

Teams whose security review says no

If your review says no to a third party holding your customers' tokens, this is built the other way: the code that reaches the network never holds a credential, and you can run it in your own infrastructure.

How it's built

Apps are just code

An App is a directory: a manifest, action files, auth methods, health checks. No SDK, no proprietary builder — the runtime types are the whole dependency.

Sandboxed execution

Every hook runs in a Deno worker with deny-by-default permissions and no ambient network — outbound traffic is proxied against a per-app allowlist.

Configuration, not deploys

Project variables, secrets and documents — a JSON key-value store — addressable from the expression editor and inlinable into any call, without touching your code.

From the blog

The newest three are below: how a Function, a Connection and a declared health check actually behave, and what arrived in the catalog lately.

Weekly Rundown: Your Own Stack, One Front Door

Sentry, Vercel, Terraform, BigQuery and 58 more infrastructure APIs landed in w6w this week — the APIs your platform team runs, beside the vendors you buy.

One Front Door: Where Your Credentials Live

Every caller outside this platform uses one front door and one credential, and here is exactly where w6w keeps the connection.

What a Function Is, and Why It Outlives a Redeploy

A Function is w6w's vendor-abstracted single operation: a stable interface that stays put while its implementation changes underneath it.

Browse the archive →

Start with the apps you already use

Reach thousands of APIs through the same front door — the ones packaged as ready-made apps, and any other HTTP API through the built-in @w6w/http. Missing one? An app is a directory of code you can import from a Git URL, so nothing here waits on a vendor's roadmap.

SlackAirtableNotionBrevoStripeLinearZendeskTodoistHubSpotGoogle SheetsBitbucketGitHubTwilioEventbriteSalesforceTrelloMailchimpContentfulOpenAIAsanaKlaviyoGoogle DriveShopifyDiscordTypeformJiraCalendlyAnthropicZoomWooCommerceIntercomGitLabGmailmonday.comDropboxPipedriveWordPressSendGridClickUpGoogle CalendarTelegramWebflowMistral AIGoogle DocsFacebook Lead Ads

Browse the apps →