Disclosure: I used an AI coding agent to collect research and draft this post. I reviewed the final copy before publishing.

I started with an Obsidian plugin idea I thought was obvious. Research killed it. A different one survived, and I'm now preselling it at $19 before writing a single line of code.

Here's the evidence that changed my mind, and the line I've drawn for myself.

The market has a reputation problem, and it's earned

Obsidian has 1,800+ community plugins and no official way to monetize them. So developers roll their own licensing. That's where things go wrong.

A real example, from a forum post about two paid AI plugins:

"Immediately after payment, I encountered a problem with the license key, which I had to request via email. While seeking help, I noticed many similar posts on the forum, indicating this issue has persisted for over a year."

A year. A paying customer waits for a license key by email, while dozens of other users report the same thing.

The CEO of Obsidian, Kepano, confirmed the company won't build an official paid-plugin marketplace: "We are explicitly not doing this." Fair enough. But it means every paid plugin author is building their own licensing, their own delivery, their own support.

Enough of them do it badly that licensing has become a recurring community complaint.

The other failure mode: ship it, abandon it

Plugins get momentum at launch and then stall. It's the most common complaint I found — "I prefer paying to have a well-maintained and feature-rich plugin rather than having to switch between abandoned ones."

That is a useful demand signal, not a guarantee. Some people will pay for maintenance, not just features.

The prices people actually pay

I collected real price points from paid Obsidian plugins:

  • SystemSculpt: $19/month or $149 lifetime
  • MeetingMind Pro: $39 lifetime
  • Typical Gumroad plugin range: $29-149
  • Top-earning Obsidian plugin on Gumroad: ~$4,200 in month one

Individual developers report $300-4,500/month as a realistic single-person range. So the "indie plugin" game is small but real. My target is at the very bottom of that range.

What I'm building, and how the research shaped it

The idea is a plugin that treats your vault like something that needs a doctor, not a chatbot. Call it a weekly checkup: orphan notes, contradictions, knowledge gaps, one report per week. The value is the report, not another chat window.

Research changed three decisions:

  1. License delivery must be part of the product. At release, payment should lead to an immediate key, not an email wait. That alone differentiates from the most established competitor.
  2. Local-first by default. Bring your own API key (DeepSeek, OpenAI, or Ollama locally). Obsidian's developer policy requires disclosing outbound requests; we disclose them in settings.
  3. Maintenance is a promise, not a footnote. Public roadmap, twelve months of updates, build-in-public from day one.

Why presell

I've been here before. My previous Obsidian plugin — obsidian-gtd, a task-management tool — hit 500 downloads and made exactly $0. It worked. People downloaded it, some used it daily. Nobody paid. A free plugin quietly teaches people your tool is worth nothing.

So this time the money comes first, or the plugin doesn't get built.

Because opinions are cheap and purchases are not. The Mom Test has a useful rule: talk about people's lives, not your idea; ask about the past, not the future. A landing page where strangers either pay or don't is the most honest interview there is.

The bar is deliberately low: two sales and I build. Four weeks with zero and I switch to the backup idea. Either outcome is data I can act on.

The presale price is $19 for the first 20 buyers (planned final price: $39). Development starts after the second presale, with delivery within two weeks after that milestone. If the second presale does not arrive within four weeks of launch, everyone gets a full refund. You can also request a refund if I miss delivery or the plugin is not right for you.

If this sounds like the tool your vault needs, here's the page. If it doesn't, tell me what you actually use for this — that's also data.

Vault Doctor — $19 presale

If the presale clears that bar, I'll post the build in public here. I plan to start with the licensing system, since that's where everyone else seems to stumble.