Run Manifestly from Claude.
Ask Claude to start a run, complete a step, reassign work, or even author a brand-new workflow from a description. Manifestly's MCP server connects to Claude.ai, Claude Desktop, and Claude Code — the assistant you already use, now driving your workflows.
mcp.manifest.lyClaude's reasoning, your runbooks.
Claude is great at multi-step reasoning, drafting structured content, and asking clarifying questions before it acts. Manifestly is where the structured work actually lives — templates, runs, assignments, due dates, audit history.
Snap them together and Claude becomes the assistant that runs your operations: from "what's overdue on the ops team this week?" to "draft me a SOC2 evidence-collection workflow with monthly cadence and import it" — without you leaving the Claude window.
- Works with Claude.ai (web), Claude Desktop, and Claude Code
- Native MCP support — no plugin, no API keys to copy
- OAuth 2.1, scoped to a single Manifestly user
- Full read, write, and author surface — not a read-only connector
- Every action attributed to the human in the audit log
Have Claude set itself up. Seriously.
Copy the prompt below, paste it into Claude, and Claude will read this page, figure out which Claude product you're using, and walk you through the exact clicks. If you don't want to read docs, this is the move.
I want to connect Manifestly Checklists to Claude using the Model Context Protocol so I can run my workflows by chatting with you. Please: 1. Read https://www.manifest.ly/integrations/mcp-server and https://www.manifest.ly/ai-integrations/claude so you understand exactly what Manifestly's MCP server does and how to add a custom MCP connector to Claude. 2. Ask me which Claude product I'm using (Claude.ai web, Claude Desktop, or Claude Code) and which plan I'm on, so you can give me the right click-by-click instructions. 3. Walk me through adding Manifestly as a custom MCP connector, using endpoint https://mcp.manifest.ly and signing in with my Manifestly account when prompted. 4. Once I confirm I'm connected, suggest three useful first prompts I can try based on what Manifestly can do (e.g. "what's overdue this week?", "start the new client onboarding workflow for Acme", "draft me an incident response workflow"). 5. Remind me at the end that everything you do via Manifestly is attributed to me in the audit log and only sees what my account is permissioned for.
This works because Claude can browse the web and read these pages itself. If your Claude tier doesn't have web browsing enabled, jump down to How to connect for the manual steps.
Three patterns Claude users keep coming back to.
Claude is particularly strong at the third one — the "design a new workflow from scratch" move. But the daily and in-the-flow patterns are where most of the value lives.
Get the read without opening Manifestly.
Ask Claude what's overdue, who's blocked, which runs are stuck. Pulls live data, scoped to your permissions. Pairs well with Claude Desktop sitting in your taskbar.
5 overdue assignments across 4 runs: • Priya (2) — Acme onboarding · 2d, 1d late • Sam (2) — Q2 SOC2 evidence · 5d, 3d late • Jordan (1) — Acme onboarding · 1d late Worst offender by total days late: Sam (8 days).
Run the workflow without leaving the chat.
Kick off runs, complete steps, set due dates, fill in fields, leave comments — everything your team does in the Manifestly UI is now a sentence in Claude. Great for Claude Code users running ops alongside engineering.
✓ Run #14228 — "Northwind onboarding" started ✓ Step 1 "Create CRM record" — completed ✓ Step 4 "Contract review" — assigned to Priya Mehta, due Fri
Design new workflows by describing them.
This is where Claude shines. Describe the SOP in plain prose — phases, owners, cadence, conditional logic — and Claude drafts the template against your account's schema, validates it, and imports it.
✓ Workflow created: "SOC2 — Monthly Evidence" Steps: 11 across 4 phases Owners: Security Lead (default), reassignable per run Schedule: Monthly, day 1 Imported into department: Security
Pick your Claude. Paste the URL.
All three flavors of Claude speak MCP and accept custom connectors. The exact path is slightly different in each.
Claude in the browser
- Open Claude.ai and go to Settings → Connectors.
- Click Add custom connector.
- Paste the endpoint URL, give it a name like Manifestly, and sign in with your Manifestly account when prompted.
https://mcp.manifest.lyClaude on your machine
- Open Claude Desktop and go to Settings → Connectors.
- Add a custom MCP connector with the endpoint URL.
- Sign in with your Manifestly account in the popup — you'll see Manifestly tools appear in the chat composer.
https://mcp.manifest.lyClaude in the terminal
- Run
claude mcp addand follow the prompts. - Use the HTTP transport with the endpoint URL.
- Authenticate via OAuth in the browser window Claude Code opens.
claude mcp add manifestly \
--transport http \
--url https://mcp.manifest.lyNeed more detail? The full MCP server reference has the capability surface, transport notes, and an admin setup guide.
Claude can only do what you can do.
We connected Manifestly to AI without giving up the boring stuff that matters: permissions, attribution, audit.
OAuth 2.1, no API keys to share.
You sign in with your Manifestly account. Anthropic never sees a long-lived secret, and you never paste one.
Scoped to your account.
Claude can only see workflows you can see, and only complete steps you'd be allowed to complete in the UI.
Every action attributed to you.
When Claude completes a step, the audit log records your name. No "AI did it" mystery actions in your history.
One click, fully revoked.
Revoke the connection from Claude's connectors panel, or from your connected apps page in Manifestly.
Not using Claude? We have you covered.
Bring Manifestly to Claude. Start free, or talk to us about your AI strategy.
With Manifestly, your team will Never Miss a Thing.
