Test results
Redirect Map Builder: test results
Tested 2026-10-08, skill version 1.0.0 at the time of loading this page. We run every skill on a strong and a weak model before it is listed, and publish both verdicts, including where the weak one fails.
Verdicts
- Date
- 2026-10-08
- Strong · claude-sonnet-5-5 (Claude Code alias "sonnet")
- Right on all 24 cases, read by hand: chains went straight to the final target and were listed under collapsed, both loops were reported with no rule for any member, old URLs with no named target stayed unmapped instead of being matched to a similar page or the home page, queries and percent-encoding were kept as written, Next.js permanent redirects and POST endpoints got 308, and both cache cases carried a note to purge or revalidate the path. The two _redirects files matched line for line.
- Weak · claude-haiku-5-5 (Claude Code alias "haiku")
- Right on 23 of 24 with the skill, but it missed one old URL: in the case with four old URLs and two unmapped, it listed only /old/news under unmapped and left /old/recipe-3 out of every list. The same case was right without the skill, so this looks like a slip, not something the skill text caused. All chains, loops, status codes and cache notes were right.
With and without the skill
Tested 2026-10-08.
| Sonnet | Haiku | |||
|---|---|---|---|---|
| with | without | with | without | |
| Redirect maps right on every check (24 cases) | 24/24 | 24/24 | 23/24 | 24/24 |
The same request on both sides; a fence around the answer is removed first. No gain on these cases for either model: Sonnet and Haiku both got 24 of 24 without the skill, including the unmapped, loop, 308 and cache-note cases. With the skill Haiku got 23 of 24 (one old URL dropped from the lists, a slip it did not make alone). The skill is a format and consistency aid here, not a measured gain, so the price is the lowest.
Same cases and the same checks with and without the skill. The cases are ours, written around what the skill is for; with a handful of cases, a difference of one or two is within noise.
Note
Twenty-four cases written by us: three chains, two unmapped lists, a two- and a three-URL loop, two cache cases worded differently, Next.js and API status codes, queries, encoded paths, a planted slug, two targets for one URL, exact case and slashes, a host change, and three _redirects files. Each answer is scored by a JSON Schema (Ajv) that accepts any rule order, either direction of a loop and 301 or 308 for a plain move, or by an exact text match for the files. Only the trap itself is strict.
What was not measured
- Models other than the two named above were not run.
- Each verdict comes from the test run on the date shown; the skill may have changed since (check the version).
- Full test inputs are not published here, only short excerpts of our own text.
- Results on your own texts, languages and domains can differ.