▸ Trust
1,019 tools are in the catalogue, last added to today. Here is exactly how they get there, how they get ranked for you, and what we do not check.
An automated discovery pass runs twice a day against public sources: GitHub, Hacker News, and — when credentials are configured — Product Hunt and a set of RSS feeds. Nobody pays to be listed and nobody can submit a paid placement.
Most of what those sources produce is news about AI rather than a tool you can use. A cheap heuristic drops article-shaped candidates before they cost anything to process — headlines phrased as “how to”, “the best”, “study finds”, names longer than eight words, and sentences. Curated listings like Product Hunt and GitHub are given more latitude than Hacker News headlines, because they are already product pages.
Anything already in the catalogue is dropped as a duplicate rather than re-added under a second name.
Each surviving candidate is scored. At 0.75 and above it publishes automatically. Between 0.4 and 0.75 it goes to a review queue instead of the catalogue. Below 0.4 it is rejected. The middle band exists so that uncertain entries wait rather than appear as though they were confirmed.
A language model writes the short description and assigns a category and audience. It does not decide whether a tool is good, and it does not set the ranking you see. When no model is configured the pipeline falls back to deterministic rules, so the catalogue never depends on a model being available.
Ranking is a scoring function over your own answers — role, goal, experience, budget, time, and current stack. That is why every pick in your stack can show the reason it was chosen. If a pick cannot explain itself, treat that as a bug and tell us.
This is the part most directories leave out. Pricing, free-tier limits, integrations and feature claims come from the tool’s own listing and are not independently confirmed. They go out of date, and vendors change plans without notice. Check the vendor’s pricing page before you commit money or data to anything here.
We also do not audit any tool’s security or privacy practices. A tool appearing in the catalogue is not an endorsement of how it handles your data.
There are none. No affiliate links, no referral parameters, no sponsored placements, no payment for inclusion or position. Links go straight to the tool. If that ever changes, it will be disclosed on this page and marked on the affected tools before it goes live — not after.
Toolnaut is in free public beta and does not currently take payment of any kind.
If a tool is mis-described, mis-priced, discontinued, or should not be listed at all, say so and it will be corrected or removed. Corrections are the mechanism that keeps the rest of this page honest.
Want to see what this produces before signing up? Look at a full example stack →