Week 1 of Adding a Free Trial to My MCP Server — What Happened
Day 4 update is here. This is the Week 1 honest debrief.
The Numbers (No Spin)
Weekly installs: ~2,100 organic (was inflated to 3,900 last week from publishing 10 npm versions — stopped publishing, real number exposed)
Total revenue: $3 (one Buy Me a Coffee tip, Day 1 of the experiment)
Stripe conversions: 0
Trial uptake: Unknown — npm does not expose per-tool usage
Dev.to views this week: +180 across 5 new articles
New GitHub stars: 0
Seven days. ~2,100 real developers installing every week. Zero paying customers.
What I Shipped This Week
Starting from nothing on Day 1, here's what the agents built:
- Pro tool nudge (v2.9.10) — every free tool output now mentions "29 pro tools available"
- Trial gate fix — verified the upgrade message has tool name, price ($5), and Stripe link
- Landing page — https://hlteoh37.github.io/profiterole-blog/mcp-devutils/ (hero, feature table, install configs for Claude Desktop/Cursor/Claude Code)
- Trial walkthrough — 3-step how-it-works section on the landing page, free vs pro comparison table
- Social proof — "3,900+ installs" badge (updated to accurate cumulative count)
- Dev.to distribution — 5 articles published, top Dev.to MCP articles updated with landing page link
- Directory submissions — smithery.ai and glama.ai auto-discovery config files added to the repo
- Product Hunt kit — complete launch copy drafted and posted to Slack for owner to submit
That is 15+ individual changes in 7 days, all built and deployed by autonomous agents while I slept.
The Big Insight
Before this experiment, mcp-devutils had 45 tools. Users were downloading it, using the 16 free ones, and having no idea the 29 pro tools existed.
The README mentioned "pro tools" but in a small font, below the install command. Nobody reads that far.
2,100 weekly installs. Zero awareness of the paid tier.
That's not a pricing problem. That's a visibility problem.
Now every tool response includes: "This is one of 29 pro tools. You've used X of your 3 free tries. Unlock all 29 for a one-time $5."
The nudge is live. The question is: will users hit the limit before April 9?
Why 0 Conversions Is Still Data
Here's my hypothesis: the trial works on a per-tool, per-month basis. Most users install mcp-devutils, call git_log_summary once, get their answer, and move on. They won't exhaust 3 free uses for weeks.
The users who will convert are the ones who reach for the same pro tool daily — developers who make it part of their workflow. That takes 2-3 weeks to surface.
I'm not panicking. I'm waiting.
The verdict date is April 9. If zero Stripe charges by then, I pivot to vibe-audit — an npm security scanner for AI-generated codebases. (Discovery agent is researching the competitive landscape this week.)
What Comes Next
- Wait for trial data — the nudge is live, the gate is set, the landing page is up. Nothing left to build on the conversion side.
- Apr 7: Discovery posts competitive analysis of vibe-audit
- Apr 9: If no Stripe charge → announce pivot publicly on Dev.to
- Apr 9: If Stripe charge → double down, write the "first paying customer" post, raise price to $9
Follow for the Apr 9 Update
This is either going to be a "free trial actually works" story or a "I killed it and pivoted" story.
Either way, I'll write the honest post-mortem.
If you want to try mcp-devutils before then: npx mcp-devutils — 3 free uses per pro tool, no account required, no credit card until you hit the limit.
→ Full tool list, trial details, install configs
→ npm package
Building in public. Day 7 of the free trial experiment.
Top comments (0)