← All runslive api
The contract
Metaintro Chat’s promise: tell us what you need — we return relevant, quality job results based on your ask, verified daily, no ghost jobs.
The claim SQA tests: The jobs returned by Metaintro Chat answer what the user asked for, judged by an LLM ensemble against the query and the user's profile. (claim C8).
This run tested the system against its contract, clause by clause. A single run can only witness some clauses; the rest stay UNKNOWN — never a faked pass.
0 pass · 0 fail · 13 unknown
- C0 MUSTUNKNOWNHeadline promisecould-not-evaluate — no relevancy score in this run
- C1 MUSTUNKNOWNUser can sign inno 'login' step in this run
- C2 MUSTUNKNOWNUser can open a new threadno 'open-thread' step in this run
- C3 SHOULDUNKNOWNOnboarding gate completesno 'onboarding' step in this run
- C4 SHOULDUNKNOWNFilters from onboarding don't bias the queryno 'clear-filters' step in this run
- C5 MUSTUNKNOWNUser-typed query is what the engine seesno 'submit-query' step in this run
- C6 MUSTUNKNOWNThe chat returns job cardsno 'wait-for-jobs' step in this run
- C7 MUSTUNKNOWNJob cards have required fieldsno 'card-shape' step in this run
- C8 MUSTUNKNOWNReturned jobs are relevant to the query · HEADLINEcould-not-evaluate — no relevancy score in this run
- C9 SHOULDUNKNOWNAll aspects of the query are coveredcould-not-evaluate — no relevancy score in this run
- C10 SHOULDUNKNOWNScore holds across rerunsneeds a sweep — a single run cannot witness this clause — needs a sweep
- C11 SHOULDUNKNOWNCompetitive vs LinkedIn / Indeed / Googleneeds a sweep — a single run cannot witness this clause — needs a sweep
- C12 MAYUNKNOWNRun completes within budgetneeds a sweep — a single run cannot witness this clause — needs a sweep
TL;DR · 30-second primer
- ·Metaintro Chat (SUT) ran 1 run on behalf of 1 seeker (P1, Lena Park).
- ·The chat returned 0 jobs. The judge scored them.
- ·Result: Job-Seeker Index 0/100. Not relevant— see “Why this verdict” (each gap maps to a claim in the Contract).
- ·Compared to 3 competitors (LinkedIn / Indeed / Google) further down.
1 ·THE VERDICT
the answer in one number30-day JSI history
METAINTRO CHAT · JSI SWEEP · RUN #27
Not relevant.
Run #27 of metaintro-chat on profile P1 for the query "deploy verification". Job-Seeker Index 0/100.
2 ·WHY THIS VERDICT
ranked by severity3 ·THE STORY
what went in, what came outInput
the same input was run against all 4 platformsQuery
Output · jobs returned
| Title | Company | Location | Posted | Link |
|---|---|---|---|---|
| No jobs returned. | ||||
4 ·THE BENCHMARK
vs. LinkedIn, Indeed, GoogleBenchmark · 4 platforms × 7 axes
- Metaintro (us)
- Indeed
Capability matrix · platforms × axes
| Platform | Recognition | Specificity | Reachability | Recency | Match quality | Hostility filter | Salary surface | Overall |
|---|---|---|---|---|---|---|---|---|
| Metaintro ·us | 65 | |||||||
| 63 | ||||||||
| Indeed | 50 | |||||||
| 56 |
5 ·SESSION RECORDING
watch the probe drive each platformSession recording
watch what the probe actually saw6 ·RUN MECHANICS
provenance & reproducibilityDuration
5.00s
Steps
—
Judges
—
Commit
—
Started
2026-05-29 03:00 UTC
Trace
Evidence by step
every artifact, link, excerpt, row, metric & recording — grouped by the step that produced itNo evidence recorded for this run.
Evidence integrity
each artifact is SHA-256 hashed at capture — proof it is unmodifiedNo integrity manifest recorded for this run.
7 ·SYSTEM ANATOMY
which component drove the verdictThis run carries no per-component probe data, so only system is shown. When a run exercises external components (Chat Engine, MongoDB, OpenRouter…), each appears here with its own health and the verdict driver is highlighted.
Press ⌘K to search