SprintScout
All within bandBands: every metric is within its 11-band envelope.
Pipeline
The Checks workflow on main: static gates, unit, component, end-to-end, deploy gate.
- Checks duration
- 6.0 min
- First-pass success
- 100 %
- Last 30 runs
- Last 30 runs on main, oldest first: 25 passed first time, 5 cancelled
- 25 first time
- 5 cancelled
Latest run on main, queue to completion
Runs on main that passed at the first attempt, of the last 30 less any cancelled
Bar height is run time. A cancelled run was superseded by a newer push.
Tests
Counts per level from the latest run; trend over the recorded runs.
- Unit
- 744 / 744 passed
- Component
- 100 / 100 passed
- End to end
- 152 / 152 passed
- Flake-free streak
- 3 runs
in 11.1 s
in 11.0 s
in 117.8 s across three browsers
Consecutive runs on main with no end-to-end failure
Intent board
Intents on main and on open pull requests, as they stood when this dashboard was built.
Built on 2026-10-31. Rebuild now
Draft 0
Nothing here
Accepted 0
Nothing here
In build 0
Nothing here
Done 61
- Target below shelter
target-below-shelter - Version in the footer
version-in-footer - Version label
version-label - Visual baselines job
visual-baselines-job - Add segment flow
add-segment-flow
+ Show 56 more
- Resolve review threads when a finding is fixed
resolve-review-threads - Versioning
versioning - Wordmark gap
wordmark-gap - Dashboard in the planner's colours
dashboard-theme - High-velocity italic logo
logo-italic - Map credit
map-credit - README banner from the real logo
readme-banner - Shelter as one segmented control
shelter-control - Unit and component badges
test-badges - Wind on/off toggle on the map
wind-toggle - Wordmark as one word
wordmark-tight - New app icon on the planner, the dashboard and the iPhone Home Screen
app-icon - Borderless Aero Stealth wordmark in the header
brand-wordmark - Run the browser jobs in Playwright's image, with timeouts
ci-playwright-image - Dashboard header in one row
dashboard-header - Start end-to-end tests alongside the other gates
e2e-parallel - Push e2e checks down to the lowest level that proves them
e2e-pushdown - Pages deploys the export that passed Checks
pages-build-once - Bring the SDLC doc back in line with the repository
sdlc-conformance - CI and dashboard hardening
ci-hardening - Cockpit fact strip
cockpit-strip - Content security policy and clear saved data
csp-clear-data - Dashboard visuals
dashboard-visuals - GPX robustness and a target that follows storage
gpx-and-target - Efforts on loop segments
loop-efforts - Name the network rule's two exceptions
network-rule - Recent runs strip
pipeline-runs - Remove the availability window code
remove-availability - Review fixes: efforts, board fetching, reopening an opportunity
review-fixes - One review per pull request
review-single - Saved data: timestamps, effort times, rider storage, GPX import, wind
saved-data - Simplify: shared helpers and unused contracts
simplify - Stale intent check
stale-intents - Test gaps and the board hour that followed a hand-picked segment
test-gaps - Aero Stealth redesign
aero-stealth - Manual rebuild of the intent board
board-rebuild - Dashboard brand
dashboard-brand - Evals on main for the dashboard
evals-on-main - Intent board on the dashboard
intent-board - Lighthouse without Lighthouse CI
lighthouse-direct - Map takes control only after a click
map-click-to-activate - Move the test mocks to msw 3
msw-3 - Remove the synthetic demo
remove-demo - Advisory gate that blocks only what a pull request changes
advisory-gate - Skill and hook evals
skill-hook-evals - Spec drafts and band issues
spec-drafts - Paid CI steps on the owner's subscription
subscription-token - End-to-end suite on a production build, with page objects
e2e-production-build - Line coverage that measures behaviour
coverage-scope - Test hardening from the Codex review
test-hardening - Deterministic bands over the dashboard history
bands - Engineering dashboard
dashboard - Evals for the agent configuration
evals - Pull request review and fix loop
review-loop - Failed-run triage
triage - Example segment on first visit
example-segment
Dropped (1)
- Efforts joined or left at a junction
junction-efforts
Quality gates
Coverage from the merged unit shards, mutation score from the last manual Stryker run, type coverage from the static gate.
- Line coverage
- 92.8 %
- Branch coverage
- 90.8 %
- Mutation score
- 81.0 %
- Type coverage
- 99.61 %
Latest manual Stryker run
Gate at 99.5 %
Performance
The deploy gate runs Lighthouse against the static export and checks client JavaScript against its ceiling.
- Lighthouse
- 100 · 98 · 100 · 91
- Perf
- A11y
- BP
- SEO
- Largest contentful paint
- 618 ms
- Total blocking time
- 25 ms
- planner client JavaScript, gzip
- 452 kB
Desktop preset on the static export
Budget 2500 ms
Budget 300 ms
Ceiling 550 kB
AI practice
Evals prove the agent configuration steers an agent; intents are the audit trail of work asked for and delivered.
- Eval pass rate
- 15 / 15
- Intents
- 62
- Review findings
- 6
- Fixed after review
- 29 %
passed · 2026-10-29 · $1.003
claude-sonnet-5-5
61 done · 1 dropped
0 blocker · 0 major · 6 minor · 0 nit over 17 pull requests
2 of 7 reviewer threads resolved, by the fix loop or by hand
Delivery
Pull requests merged into main, read from the GitHub API by the dashboard job.
Median from opened to merged
This week; the chart shows the last eight
Slowest tests
The five slowest end-to-end tests in the latest run, across all three browser projects.
| Test | Project | Duration |
|---|---|---|
| ranking fetches one forecast per segment and puts the gain first @smoke @a11y | webkit-a11y | 22.7 s |
| efforts panel is keyboard reachable and described | webkit-a11y | 18.5 s |
| Enter on an hourly row selects that hour @keyboard @a11y | webkit-a11y | 17.8 s |
| terrain panel lists the sampled elevations | webkit-a11y | 17.1 s |
| a preset can be chosen with the keyboard @keyboard | webkit-a11y | 16.5 s |