Show HN: I built a linter for landing page conversion

[nikhonit]

Hey HN,

I'm Nik and I've worked 8 years working in Marketing at High Growth Startups. I’ve spent years fixing landing pages for my campaigns, and I realized most bad conversion rates come from structural errors, not just bad copy. (Still copy remains the major part of conversions)

I built Landkit Audit to act as a Linter for conversion.

How it works:

Ingestion: It fetches your site and converts the DOM into a high-fidelity semantic map. This strips away the design noise to reveal the actual information hierarchy your user is navigating.

The Linting Rules: It runs this structure against 50+ heuristics derived from direct response frameworks (Cialdini, Fogg, LIFT).

Heuristic Analysis: It flags specific 'violations' like:

Visual Hierarchy: H1s that fail to pass the 5-second test (inferred).

Friction: Forms that ask for high-commitment data before establishing value.

Anxiety: Missing trust signals near payment triggers.

The Stack:

Frontend: React + Framer Motion (for the Terminal visualization).

Backend: Supabase Edge Functions.

Analysis: Gemini Pro 2.5 (great at logical reasoning/inference).

The hardest part was tuning the prompt to be critical rather than nice. Most LLMs default to being supportive; I had to prompt engineer a persona that acts more like a brutal senior editor.

Try it here: https://landkit.pro/audit

It’s free to run the audit. I’d love to know if the Linting errors it finds on your projects feel accurate.

[nikhonit]

Note on speed: The audit takes about 15-20 seconds to run.

Since we are spinning up a headless browser to map the DOM before feeding it to the inference engine, it's not instantaneous like a regex check