Corsair

The open source integration layer for AI agents.

Visit website

SubscriptionAIAPIAI Agents2 impressionslaunches Nov 201 comment

>log in to comment

Dev Jain[maker]· 21d ago
We built Corsair to give AI products a simple, secure way to add multi‑tenant integrations—OAuth, permissions, webhooks and hosted or self‑hosted execution—without managing the infra. It lets you connect apps like Gmail, Slack, Notion in minutes. Would love your thoughts on the experience 😊

Corsair is an open source integration layer for developers building AI agents and applications. Modern AI agents need to connect to dozens of external services — GitHub, Slack, Gmail, Google Drive, and 200+ others but every provider comes with its own OAuth flow, token refresh logic, rate limits, and webhook format. Corsair handles all of that once, exposing every integration through one consistent syntax. Built natively on the Model Context Protocol (MCP), it plugs directly into AI agent frameworks and can be exposed as an MCP server endpoint for any MCP-compatible client. Crucially, Corsair runs inside your own application and never stores user credentials on its own infrastructure everything stays encrypted in your own database, while the hosted Hub only manages connect pages and OAuth callback URLs. It's built for developers and teams shipping AI agents, internal tools, or SaaS products that need reliable, secure connections to third-party APIs without reinventing integration plumbing for every new provider.

AI API AI Agents

Corsair is an open-source integration layer that lets AI agents and apps connect to 200+ third-party services via a unified API.

for: Developers and teams building AI agents, internal tools, or SaaS products that need secure third-party API connections.
pricing: freemium · free trial
license: open source

corsairdev/corsair 13k 671TypeScriptupdated 21 days ago

v0.1.0a year ago

works withGmailSlackNotionGitHubAirtableStripeHubSpotLinearDiscordGoogle DriveSheets

Key features

6 features of Corsair

Unified integration syntax — Expose all supported services through a single, consistent API regardless of each provider's OAuth flow.
Self-hosted SDK — Run the full integration stack on your own infrastructure with no per-seat fees.
Managed permissions — Set permission modes per integration; destructive actions require explicit user approval.
Caching & rate-limit handling — Data is cached via webhooks and polling, reducing API calls and handling rate limits automatically.
Open source plugins — Add or customize integrations by opening a PR or forking the TypeScript repo.
MCP compatibility — Plugs directly into Model Context Protocol-compatible agent frameworks or can be exposed as an MCP server.

Use cases

Add a "Create calendar invite" action to a SaaS product without writing OAuth code.
Sync Airtable records to your internal database while handling token refresh automatically.
Trigger a Slack notification when a file is uploaded to Google Drive.
Allow users to send emails via Gmail from within an AI-driven chat interface.
Build a multi-tenant AI assistant that accesses each tenant's third-party credentials securely.

Corsair vs alternatives

all alternatives →

	Corsair	Langfuse	QueryPal	WunderGraph
Best for	Open-source integration layer for AI agents	LLM observability	Support automation	API composition
Pricing	Freemium	Subscription	Free	Free
DevHunt upvotes	1	89	82	38
Launched	Nov 2029	Jan 2023	Mar 2024	Jan 2023

Corsair vs Langfuse: Langfuse provides observability for LLM apps, not integration plumbing.
Corsair vs QueryPal: QueryPal automates support team workflows, whereas Corsair focuses on generic third-party API integration.
Corsair vs WunderGraph: WunderGraph composes APIs into a unified GraphQL layer, while Corsair provides a unified OAuth and webhook integration layer.