Claude + Heroku: What the Integration Can (and Can't) Do in 2026
Yes — Heroku ships an official remote MCP server you can add to Claude. Be precise about the wording, though: Heroku is not in Anthropic’s connectors directory (claude.com/connectors/heroku returns a 404). What exists is an official remote MCP server at https://mcp.heroku.com/mcp, announced June 2025, that you add as a custom connector. Wire it up and Claude can manage app lifecycle, browse add-ons, and read team and Private Spaces info from a chat. The catch is the usual one: it only works inside a conversation you start. There are no triggers, nothing watches your apps, and nothing happens while you’re away.
Here’s exactly what the integration does, how to turn it on, where the limits bite, and what to use if you want Heroku work that runs on its own.
What the Heroku MCP server does
The server connects Claude to your Heroku account and exposes platform operations as tools you drive in natural language.
In practice, Claude can:
- Manage app lifecycle — list apps, read app info, create and update apps.
- Read teams and Private Spaces — pull team membership and Private Spaces details.
- Browse the add-on marketplace — look up and reason about available add-ons.
- Do more locally — the official local server adds database (
pg:*) operations, scaling, add-ons, and logs on top.
Heroku’s own announcement name-checks Claude (“from your favorite agents such as Claude, Agentforce, or Cursor”), so this is a supported, first-party server — just not a one-click Anthropic directory entry.
How to set it up
The remote server is OAuth-based, so it slots in as a custom connector:
- In Claude, open Settings → Connectors → Add custom connector and point it at
https://mcp.heroku.com/mcp(remote custom connectors on claude.ai require a paid plan). - Authorize through the id.heroku.com browser flow — it’s OAuth 2.0 with token refresh. Your client needs to speak Streamable HTTP (not SSE); Claude Desktop historically needed the
npx mcp-remote https://mcp.heroku.com/mcpproxy. - Back in a chat, ask Claude to list or create apps; it hits the MCP server on your behalf.
For a local setup, the official server runs via heroku mcp:start (Heroku CLI v10.8.1+, reusing your CLI auth) or npx -y @heroku/mcp-server with a HEROKU_API_KEY. One caveat: Claude Enterprise/Team plans may restrict adding the URL — admin permission can be required.
The limits that actually matter
The server is genuinely good at pulling Heroku into a conversation. But its shape is “an operator you direct,” not “an agent that runs.” Three limits define it:
- No triggers, no monitoring. The server only works inside a conversation you start. There’s no “restart the dyno if it crashes” or “alert me when a release fails.” Heroku’s Platform API supports app webhooks (release, build, dyno, add-on events), but the MCP server itself doesn’t act on them — you have to be there, prompting.
- Conversation-only. Claude manages an app when you ask; it doesn’t sit watching your releases and reacting on its own. Close the chat and nothing continues.
- Laptop-bound for anything scheduled. The closest thing to “running on its own” is Claude Cowork’s scheduled tasks, which fire on a fixed clock, not on inbox events. That’s not an always-on, event-driven ops agent.
Worth separating: the Heroku MCP server (agents control Heroku) is a different thing from Heroku Managed Inference (Heroku hosting models — including Claude Opus 4.5 — for your apps). Don’t blend them. Claude via the MCP server is great for “create this app and check its config” — not for “watch the fleet and act when something breaks.”
If you want Heroku work that runs on its own: Carly
The moment you want something to happen around Heroku without you in the chat — email the team when a release fails, log an incident when a dyno crashes, act on a build event and route the result — you’ve crossed past what Claude’s MCP server is for.
That’s where Carly fits. Carly is an AI executive assistant built to act on triggers, not just answer in a chat:
- Fires on events, 24/7, in the cloud — hang a workflow off Heroku’s app webhooks (release, build, dyno, add-on events POST to your URL); your laptop doesn’t need to be awake.
- Connects Heroku to the rest of your stack — take a failed release or a crash event and turn it into an email, a task, a CRM note, or a calendar block in the same workflow.
- Actually sends and updates — drafts and sends email (Gmail and Outlook) with attachments, files and labels, manages tasks, updates your CRM, and records meetings.
- Builds the workflow for you — tell it “when a Heroku release fails, email me the details and open a task” in plain English; it interviews you, then builds it with you. No prompt engineering.
AI agents start at $35/month, and steps in a workflow that don’t use AI run free and unlimited. Carly connects to 200+ tools across 40+ categories — see integrations. Heroku connects to Carly via your own API key — paste your Heroku API key on carlyassistant.com/integrations, and Carly can do whatever Heroku’s Platform API allows.
Claude vs Carly
| Claude (Heroku MCP) | Carly | |
|---|---|---|
| Manage app lifecycle | Yes | Yes |
| Browse add-ons | Yes | Yes |
| Acts on triggers / events | No | Yes |
| Reacts to a failed release on its own | No | Yes |
| Works while laptop is closed | No | Yes (cloud) |
| Sends notifications as email | No (Gmail draft-only) | Yes (Gmail + Outlook) |
| Pricing | Pro $20 / Max $100–$200 | AI agents from $35/mo |
Claude’s MCP server is a strong Heroku control panel inside a chat. Carly is a teammate that reacts to releases and crashes on its own.
Frequently Asked Questions
Does Claude work with Heroku?
Yes, through Heroku’s official remote MCP server — but it’s a vendor MCP you add as a custom connector, not an entry in Anthropic’s directory (claude.com/connectors/heroku is a 404). Point Claude at https://mcp.heroku.com/mcp, authorize over OAuth, and Claude can manage app lifecycle, browse add-ons, and read team info inside a chat.
How do I connect Claude to Heroku?
In Claude, go to Settings → Connectors → Add custom connector and enter https://mcp.heroku.com/mcp, then authorize through the id.heroku.com OAuth flow (Streamable HTTP required; Claude Desktop may need the mcp-remote proxy). For a local setup, run heroku mcp:start or npx -y @heroku/mcp-server with a HEROKU_API_KEY. Enterprise/Team plans may require admin permission to add the URL.
Can Claude react to Heroku events automatically?
No. The MCP server works inside a conversation you start — it doesn’t act when a release fails or a dyno crashes. Heroku’s Platform API supports app webhooks, but you need an agent platform like Carly to catch them and act 24/7 in the cloud.
Is this the same as Heroku Managed Inference?
No. Managed Inference (Heroku AI, now GA) hosts models — including Claude Opus 4.5 — for your own apps to call. The MCP server is the reverse: your Claude agent controlling your Heroku account. This post is about the MCP server.
More: Claude connectors · Claude + GitHub · Can Claude send emails · Claude vs Carly · Claude Cowork alternatives · Best AI tools for solopreneurs
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
