Ask This Before You Migrate Your PMIS

I spent five minutes finding my new project management tool’s real limit. Not a marketing limit — the actual wall. I was migrating 27 projects, 40 sections, 232 tasks, 25 labels, and 15 notes into ClickUp, and by the time the batch job finished tagging everything, ClickUp’s free tier had already stopped answering.

This isn’t an argument that every operator needs to bring their own AI model to their project management tool. Asana ships a genuinely capable agent now. Monday’s Sidekick builds entire apps inside the platform. If either of those covers what an operator needs, nothing here applies. This is for the narrower case: the operator who wants their own model — Claude, in my case — doing the work inside the tool, not the vendor’s. For that operator, here’s what actually matters before migrating, because I didn’t check it first and it cost me real time.

The Free Tier Ran Out in Five Minutes

ClickUp’s free tier allows 50 API calls in a rolling 24-hour window — not a fixed daily reset, a rolling one. The $7/month Unlimited plan raises that to 300. I hit the ceiling inside a single batch-build session, and because the window rolls instead of resetting at midnight, the fix wasn’t “wait until tomorrow.” It was “wait up to a full day from the exact moment I tripped it.” Every other tool I checked throttles per minute, not per day — Asana, monday.com, and Wrike all recover in under sixty seconds. ClickUp is the only one where a single burst costs an afternoon, sometimes more.

Two Seats Cost What One Solo Tool Charges

The number that actually stops a solo operator isn’t the API limit — it’s the seat count. Wrike’s Team plan won’t sell to fewer than two seats, even to a one-person account — $20 a month at $10 a seat. That gets Gantt charts. It doesn’t get critical path: Wrike walls that off until Business, which has its own floor — five seats minimum, $125 a month, still for one person. The seat floor doesn’t shrink for a solo operator. It just prices the feature further out of reach the more of it there is to pay for.

ClickUp — the same tool whose free tier had already stopped answering me — turns out to price this better. Its Business plan has no seat minimum at all, and $19 a month gets a single account both Gantt charts and full critical path: less than Wrike’s two-seat floor, for the feature Wrike’s two-seat floor doesn’t even include. The sticker price on the pricing page isn’t the real price. The real price is what the feature actually costs once the seat floor underneath it gets factored in — and that floor isn’t the same shape on every platform.

The API Isn’t the Real Limit — the MCP Is

Here’s the check almost nobody runs: a platform’s advertised API limit and its MCP server’s limit are not the same number. Asana’s MCP server enforces its own, tighter ceiling, separate from what the REST API allows. Evaluate a tool by its API docs alone and that’s pricing the wrong door — the one an agent actually walks through is often narrower than the one in the marketing copy.

I Built a Router Because the Platform Wouldn’t

I use both of Asana’s doors — the MCP and the raw API — because neither one alone covers everything I need at the throughput I need it. So I built a skill that decides, per operation, which one to use: the MCP where it’s the right tool for the job, the API where the MCP’s ceiling would choke a batch. I didn’t design that routing logic in advance. I found it by hitting the wall on one side, switching to the other, and doing that enough times that the pattern held.

Todoist Had the Budget, Not the Chart

Todoist’s limits are the most generous of anything I checked — 1,000 requests every 15 minutes on the REST API, enough headroom that rate limiting was never a real concern. I ruled it out anyway. As a PMP-certified operator, I need to forecast time, effort, and duration against a real critical path, and Todoist has no Gantt view to show me one. I tried bolting on a Ganttify plugin; the UI was overwhelming enough that I stopped using it inside a week, and it doesn’t connect to Claude at all. A generous API attached to the wrong tool is still the wrong tool.

What to Check Before Migrating

Four questions, in the order I wish I’d asked them:

Check	Why it matters
What’s the MCP limit, separate from the API limit?	An agent can hit a tighter wall than the docs advertise.
Does the feature in question require a second seat?	Some platforms don’t sell a one-seat plan for it at all.
Is the throttle per-minute or per-day?	A rolling daily window turns one burst into a lost afternoon.
Does it have the actual PM structure required — Gantt, critical path, dependencies?	A generous API on the wrong tool doesn’t fix the tool.

I didn’t ask any of these before I started migrating. I’m asking them now so the next person doesn’t have to find out the same way I did.

I Don’t Know Yet — And That’s the Honest Answer

I’m currently on Asana’s 14-day trial, and I’m still weighing whether to build my own PMIS on top of Airtable instead. I don’t have a tidy ending yet — I’ll know more in about two weeks. What I have is a working list of what to check, and a real cost, in hours, for not checking it the first time. That’s the trade I’m offering: skip my five minutes with ClickUp, and spend them somewhere that actually pays off.

Sources

• ClickUp's MCP server rate limits, straight from their own docs — ClickUp Developer Docs
• ClickUp's plan comparison table — Business plan seat minimum (none) and Gantt/Critical Path & Slack Time availability by tier — ClickUp Pricing
• Asana's per-token rate limits and `Retry-After` behavior — Asana Developers
• Todoist's REST and Sync API limits — Todoist Developer Docs
• monday.com's per-minute and daily call limits — monday.com Developer Docs
• Wrike's REST API throttling behavior — Wrike Help Center
• Wrike's plan pricing and seat minimums (Team: 2–15 users; Business: 5–200 users) — Wrike Pricing
• Wrike's critical path feature, confirmed available only on Business and higher — Wrike Help Center: Critical Path