Observatory.
Real work, seen from a different angle.
Every part
of the story↗
What we plan. What we protect. What our machines actually do.
From intention
to done.
Workspace activity across the explicitly scoped teams.
505open issues
One full square = 4 issues; dim squares show a remainder. Current non-archived work at capture time.
Making things happen.
Completed work, by team.
Issue counts describe deliveries in Linear. Team estimates use different scales and are not compared here.
The cycle’s shape.
Coverage & counting
Engineering issue, project and cycle aggregates. Current backlog is a present-day census; event counts use recorded timestamps.
Scope: CEP · CODOS · DES · FB · FINE · FRTNS · PAY · RET. Current work stays current when the period changes. Historical totals use each issue’s latest timestamps; reopened and repeated transitions are not reconstructed as an event log.
Captured project census: 80 projects. Lead time median: 8.9h from 2119 completed issues with timestamps.
The city’s
early warning.
Captured error traffic and issue health across the configured scope.
Error traffic.
Accepted-event totals: 04 Aug, 16:00 — 02 Nov, 17:00 UTC · hour-aligned bounds. The current hour may be partial. Traffic outside this captured interval is unknown.
125issues seen in this range
These are the latest statuses of issues seen in the selected range, not a reconstructed history of triage transitions.
Across the services.
Coverage & counting
Issue, release and accepted/dropped event aggregates. Counts reflect source retention and the selected scope.
Organization-wide project scope. Issue pagination: complete. Releases: complete. Event volume is accepted error traffic queried on UTC hour-aligned bounds, shown above. Issue and release timestamps use the requested period. Dropped traffic is a separate ingestion outcome.
Inside the
intelligence.
Observed work, model usage and the tools behind the runs.
7,353captured traces
Estimated cost uses telemetry pricing, not a provider invoice. Sessions and tool activity have separate source cohorts; usage totals are not added together.
Where model span cost goes.
Span cost and span counts form their own cohort. Trace usage above uses the trace aggregation.
Agent Score.
No score published for Nov 2.
Latitude requires enough evidence in every dimension. An earlier score is not substituted for today.
Calls, failures and waiting.
Coverage & counting
Engineering traces from the primary project's 90-day cohort; 2 project(s) are available separately in telemetry. Sessions, tool calls and spans are separate cohorts; cost is estimated.
Project: Fortis Agents. Requested window: 04 Aug, 16:48 — 02 Nov, 16:48 UTC. First ingested trace: 25 Sept, 09:14 UTC. Source retention and uncaptured work cannot be reconstructed.
Real sources. Explicit coverage. No invented work, no inferred human-only share, no prompts or customer payloads.
Inspect the sourcesKeeping the
engines running.
Production and agent services. Captured capacity and resource signals, with room for the unknown.
15 observed services across 2 configured clusters · current at capture time.
Means give each captured service-hour equal weight. They do not represent resource-weighted fleet load.
Aug 4 — Nov 2 UTC · metric query windowA closer look at the fleet.
media-preview-delivery
fortuneos-prod · active
Inspect hourly values
| Hour | CPU | Memory |
|---|---|---|
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 1.0% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 1.0% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 0.9% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| 1.0% | 3.5% | |
| Not captured | Not captured |
On the engine now.
Deployment objects present at capture time, independent of the page period.
- Running / desired
- 1 / 1
- Pending
- 0
- Failed task attempts
- 0
These objects are not a release history. Failed task attempts are the provider’s current deployment counter.
Coverage & counting
Current ECS service inventory and observed service-hour CPU/memory aggregates. Missing metric buckets are unknown.
Scope: fortuneos-prod · fortuneos-prod-agents-services · eu-central-1. Desired, running and pending task counts describe capacity at capture time. Reaching the desired count does not establish application health.
CPU and memory headline ranges follow the selected 90-day window. The monitor always shows the last 24 hours of hourly averages. Minima and maxima come from the provider’s hourly minimum and maximum statistics; missing data remains unknown.
Collection: 4 ECS reads · 4 CloudWatch reads. No task environment, secret values or application request data are included.