https://agentopt.app · status
System status
Public health snapshot of the curated agent catalog and select experience. Updated from automated Observer scorecards (typically daily).
System health includes traffic and probe gates beyond catalog size. Catalog fill may still be on target. Operators: full scorecard on local Analyst /ops.
Active and total candidates are live catalog counts (same as Info). Latency and other health metrics are from the latest Observer scorecard as of 2026-09-30T13:03:09.
Catalog
Active (45 days)
5134
Total candidates
8732
Skills breadth
11449
Skills stable
1084
Skills assignments
80565
Hapax share
80%
Skills are raw distinct / stable (n≥3) / assignments / hapax share (latest Observer scorecard).
Re-confirmed (14 days)
96%
Active is the 45-day window of the live catalog. Total includes older and archived rows. The 14-day share is how recently those actives were re-checked — not a quality rating of the tools.
Select experience
These figures are outcomes of a daily synthetic standardized probe of production select (fixed query suite, free and paid lanes) — not organic customer traffic.
Success rate
100.0%
Checklist relevance
72.2%
Share of 18 standard cases whose top results match the expected capability family, or a name anchor when the catalog row has no capability tags, and that also satisfy the case's constraints. Free and paid use this same judgment. A free response is a shorter list and omits detail fields. When the paid request does not set steering attributes such as cost, maturity, or reliability, that shorter list is the leading part of the paid list, in the same order. A paid request that sets those attributes may rank differently. A result's own fit is its score: 0.55 usable, 0.72 or higher strong. The 0.85 line is an internal warning, not a service outage.
Paid richness · 4.8 / 5
Paid select includes extra fields on the top-ranked candidate that free select omits: capability tags, quality dimensions, score breakdown, clarifications, and a source URL. The score is how many of those five fields were present on average in the latest synthetic paid probe — not organic traffic. Typical healthy paid responses are 4–5.
Typical latency band · latest scorecard
Typical free-tier select latency about 1700–7000 ms
Typical paid-tier select latency about 1200–2000 ms
Confirmed range from the last automated health check — not a multi-day latency chart.
Quality & bar-raise
Typical maturity
0.90 / 1 · very high
Typical reliability
0.88 / 1 · very high
Typical first-pick match
0.74 / 1 · strong
Bar-raise · improving · ↑ 10 / ↓ 4
Typical maturity — Typical catalog maturity on a 0–1 scale (higher = more established): how established the typical active candidate looks. Below 0.35 is low, around 0.55 is mid, 0.75+ is very high. The number is the median — half of actives sit at or above it.
Typical reliability — Typical catalog reliability on a 0–1 scale (higher = more dependable): how dependable the typical active candidate looks. Below 0.35 is low, around 0.55 is mid, 0.75+ is very high. The number is the median — half of actives sit at or above it.
Typical first-pick match — Typical first-pick match: average 0–1 match score of the #1 candidate on the daily synthetic probe (same scale as each select score). 0.55 is a usable first pick; 0.72+ is a strong one. Higher means the standard queries found a stronger first choice. Not organic traffic.
Catalog trend
| As of | Candidates (active/total) | Skills (breadth/stable/assignments/hapax share) | Catalog health |
|---|---|---|---|
| 2026-09-30T13:03:09 | 5134 / 8732 | 11449 / 1084 / 80565 / 80% | healthy |
| 2026-09-29T13:03:16 | 5122 / 8683 | 11420 / 1107 / 79668 / 80% | healthy |
| 2026-09-28T13:03:08 | 5103 / 8624 | 11436 / 1100 / 79432 / 80% | healthy |
| 2026-09-27T13:03:04 | 5094 / 8575 | 11191 / 1059 / 78772 / 80% | healthy |
| 2026-09-26T13:02:47 | 5089 / 8517 | 11334 / 1061 / 77576 / 81% | healthy |
| 2026-09-25T13:02:41 | 5097 / 8450 | 11141 / 1021 / 75395 / 81% | healthy |
| 2026-09-24T13:02:47 | 5094 / 8349 | 10703 / 962 / 71481 / 81% | healthy |
Catalog size, skills, and catalog-local health only (not system traffic/probe gates). Skills are raw distinct / stable (n≥3) / assignments / hapax share. Machine JSON: https://agentopt.app/v1/status