The Wire Test is published by LABS, the studio that builds labs.llc. Every tool in this log was built by the people running the tests — how the log guards against that.
Method
How a run is logged
A Wire Test log is a record of what a tool did when we asked it to work, written down as we went. This page covers the bench we test on, what each row of a run table means, how rivals are picked and read, and what we could not test.
Ownership
What owning the tools means for our verdicts
We are marking our own work, and a reader should discount a maker's praise. So we hold ourselves to rules that make the logs checkable rather than persuasive.
- No ratings, no stars and no review markup for search engines. A number from us about our own tool would mean nothing.
- Every log names at least one real fault; most name four or five. They sit in their own leg, not in a footnote.
- For every rival, what it does better comes first.
- Claims about a labs.llc tool cite a file and line in its source or a timed probe of its local copy. Claims about a rival come from its own page, read that day.
- Rivals are linked plainly; we show no rival logos or screenshots and quote no prices. The link to LABS in the footer is marked nofollow.
The bench
Logged here too: each page view on the.net.im goes to Google Analytics 4, under this site’s own property rather than labs.llc’s, and that tag keeps its _ga cookies in your browser.
What we test against
Every run tests build 537 of labs.llc: its source tree, which is byte-identical to the live site, and a local server that serves that tree. We did not load labs.llc itself over the internet for these runs. Where a tool calls an outside source — Treasury, the SEC, the FAA, JPL — we called the relay or endpoint the page calls, at the time shown, with the page's own parameters unless the row says otherwise. Where a tool ships an engine, we cut it out of the page and ran it in JavaScriptCore.
The screenshots are of labs.llc tools only, taken from the same local copy.
A run
A run, step by step
- Read the tool's source and note the files, lines and endpoints it depends on.
- Load the page from the local copy and record its status and size.
- Call what the page calls — relays, public APIs, shipped data — and write down what came back.
- Try to misuse any relay the tool runs: another address, a bad parameter, an internal host.
- Read two or three rival pages and note what each offers, with the time.
- Write the findings, then the faults, then the sign-off.
Rows
What a row records
A row is one probe: its time in UTC, what we asked, and what came back — the HTTP status and size in bytes where there was one and, where we measured it, how long it took. Our own probes are stamped to the second; rival pages were logged to the minute, and the table shows that difference rather than inventing seconds. A row marked not run is something we chose not to do, with the reason beside it.
Stamps are written in UTC. The Times in switch converts them to your own time zone inside your browser; nothing is sent anywhere, and the choice is remembered on your device only.
Rivals
How rivals are chosen and read
A rival is a real, current tool that a reader might use instead: an agency's own page, an open-source program, a commercial service. We read each one on 27 September 2026 between 19:59 and 20:12 UTC. When a rival's page would not load for us, we say so and claim only what we could see: CourtListener's site refused our fetcher, so its log leans on the Free Law Project's own description; the FAA's status dashboard draws itself by script, so its log claims no more than its page titles. We never say a rival cannot do something unless we checked.
Limits
What this session could not test
- Backgammon's two-device relay: opening a game writes a file into the build tree, which we may not change.
- Portal's framing headers: they live in the site's .htaccess, which the local server does not apply.
- Docket's full-text relay on the live site: the tree holds no token, so we cannot say whether live has one.
- Recall's FSIS path: it only answers browsers, and we did not run the page in one for this log.
Sign-off
The sign-off, not a score
Each log closes with four lines: what held up under test, what fell short, when to reach for the tool, and when to look elsewhere — usually at one of the rivals by name. If you find a row that is wrong, write to labs@labs.llc and name the page and the row.