Connect 5DAYS to Claude, ChatGPT, and Gemini via MCP

Your team’s knowledge lives in 5DAYS. Your AI habits might live elsewhere — Claude for writing, ChatGPT for brainstorming, Gemini inside your browser. As of today, they can meet: the 5DAYS MCP server is live for all plans.

MCP — the Model Context Protocol — is the open standard that lets AI assistants connect to external tools and data sources. Add 5DAYS as an MCP server, and the assistant you already use can search your notes, read your transcripts, and check your tasks. Your workspace becomes context, wherever you’re asking questions.

What it looks like

Once connected, you can ask your assistant things it previously had no way to know:

• “Search my 5DAYS workspace for what we decided about the Q3 roadmap.”
• “List my open tasks from the client onboarding project.”
• “Pull the transcript of yesterday’s standup and summarize the blockers.”

The assistant calls the 5DAYS MCP server, the server answers with real data from your workspace, and the response cites what it used. No copy-pasting notes into chat windows, no stale exports.

Permissions are the whole point

An integration that bypasses access control isn’t a feature — it’s an incident waiting for a date. So the rule is simple: the MCP server enforces exactly the permissions of the person who connected it.

1. You authenticate with your own 5DAYS account.
2. Every request the assistant makes runs under your access rights.
3. If you can’t open a project in 5DAYS, no assistant connected as you can read it either.

The same applies to our public API, which the MCP server is built on: GET /api/v2/search returns what you can see, nothing more. Connecting an external AI never widens access — it only changes where you ask from.

Setup in two minutes

For Claude, ChatGPT, Gemini, or any MCP-compatible client:

1. In 5DAYS, go to Settings → Integrations → MCP server and generate a connection token.
2. Add the 5DAYS server in your assistant’s MCP or connector settings, and paste the token.
3. Ask something only your workspace would know. Check the sources.

Revoking is one click, tokens are scoped and expiring, and every MCP request appears in your workspace audit log — so your admin can see exactly which external tools are connected and what they touched.

An open ecosystem, deliberately

We build Spark AI to be the best way to reason over your workspace — it runs where your data runs, on European infrastructure. But teams are pragmatic, and locked doors don’t make products better. If your workflow lives in another assistant, your knowledge should still work for you there, safely.

The MCP server is included in all plans, starting today. Documentation and the full tool list are in the developer docs — and if you build something on it, tell us. We’re curious.