How We Rate
Every directory item and app page shows an editor's rating from 0 to 5. These are editorial judgments by the YourAIKit team, scored against the rubric below. They are not user reviews.
The Criteria
Each criterion gets a score from 0 to 5 in half points. The editor's rating is the weighted average, rounded to one decimal.
| Criterion | Weight | What we look at |
|---|---|---|
| Usefulness for real work | 35% | Does it produce something a working professional sends, files, or acts on? Everyday tasks score higher than niche or novelty ones. For apps: how well it serves the work tasks on this site. |
| Quality and clarity | 25% | Are the instructions or docs specific, structured, and clear about the output? We read every hosted copy in full. Short outlines and loose link lists score lower. |
| Ease of setup | 15% | How fast can a non-developer get a first result? Paste-and-go prompts score highest. Skills that need connected apps, paid partner accounts, or code and API keys score lower. |
| Maintenance and activity | 15% | Is it kept current? For GitHub items we check the last commit to the item's own file or folder and to the repo, plus star counts and README notices, fetched live on the day we rate. Archived or retiring items score lowest. For apps: active development and documented agent access. |
| License and trust | 10% | Is the license clear and the publisher accountable? An open license from a known publisher scores highest. No stated license, or an opaque closed listing, scores lower. |
What the Numbers Mean
| Rating | Meaning | Entries now |
|---|---|---|
| 4.5 to 5.0 | Exceptional. Useful most weeks, well made, current, and clearly licensed. | 11 |
| 4.0 to 4.4 | Strong. A good pick with a small catch, such as setup or a narrower audience. | 130 |
| 3.5 to 3.9 | Solid, with a real caveat: thin instructions, connected apps required, or a narrower job. | 170 |
| 3.0 to 3.4 | Useful in a narrow case. Look at alternatives first. | 53 |
| Below 3.0 | Hard to recommend now: archived, being retired, moved, or better done elsewhere. | 1 |
365 entries rated on Oct 17, 2026.
How We Gather Evidence
- We read every copy hosted on this site in full, and we open the source page for everything else.
- For GitHub items we fetch the star count, the last commit to the repo, and the last commit to the item's own file or folder on the day we rate, and we check the README for archive or retirement notices.
- We check that every source link still works. A link that redirects somewhere else, or a product the vendor is retiring, scores lower on maintenance.
- We keep a short evidence note and the date rated for every score. The full set is published at /ratings.json.
Apps
Apps are rated on how well they serve the work tasks on this site: the kits, skills, and prompts in the directory. That is not a full product review. Where StackFinds has published a review, the app page links to it.
Legal Items
Legal prompts and skills are rated as drafting and review aids for lawyers and compliance teams. A high rating means the item is a useful starting point for a lawyer, not that its output is correct. Nothing here is legal advice.
No Paid Influence
Nobody pays for a rating, and nobody can buy a change to one. Sponsored or featured items never get a rating boost. If we add sponsorship, it will be labeled on its own and kept separate from the editor's rating. Our own paid packs are not rated here.
We do not mark editor's ratings up as reviews in search-engine structured data. They are shown to readers only.
User Ratings
User ratings, when they arrive, will be shown separately from the editor's rating. The two are never averaged together.
Corrections
If a rating rests on something out of date, such as a project that has moved or a license that changed, tell us and we will re-check it.
YourAIKit products are workflow tools. They are not legal, tax, or financial advice.