MOJI
Inside thresholds $5,020 / $6,180 · wk 33
RD

Overview

99.2%
Reliability
▲ +0.3 pt vs last week
runs inside the committed budget · 8 weeks
wk 26wk 33
This week34,339 of 34,616 inside
Stopped at the ceiling, state kept6 · all resumed
Worst week (26)98.4%
34,616
Runs
▲ +12.4%
$5,020 of $6,180
Spend
$1,160 headroom
daily spend against the week's ceiling
ceiling pace MonSun
Biggest lineinvestigations · $1,290
Biggest stepreach a verdict · $542
Headroom kept$1,160
94.8%
Yield · selected
▲ +1.1 pt
by workflow, this week
security · 7 lines96.2%
support-resolutions96.1%
repo-tasks88.4% · below the bar
Weakest linecontainment-actions · 92.8%
3m 40s
p95 cycle
▼ −18s
2 open
Exceptions
1 escalating
13:40recovered in line inv-7f3a2c · step 4 13:39held at ceiling inv-79c2e1 · ~$0.06 finishes 13:38commit v3 investigations · $1,600 wk
Live →

Spend

by workflow · wk 33
security7 lines $2,540 / $3,120
support-resolutions $1,660 / $2,180
repo-tasks $820 / $880

6 stops · state kept · resumed

Yield

daily · wk 33
98% 95% 92% quality threshold · 92% Thu: 95.1% passed first time 6 re-runs, all recovered in line 94.8% Mon Thu Sun

Check scores

verdict · 4,182 runs
threshold 0.80 mean 0.91 0.60 0.85 1.00 re-run cluster

Proposals

Promote support-resolutions to autonomy?
Earned, not switched on: 4 of 4 weeks inside every threshold, first-pass yield 96.1% and rising, no ceiling breach. The review queue has had nothing to catch for a fortnight.
Move the correlate step to the small model?
On 500 shadowed runs the small model held 0.91 quality against your 0.80 threshold at an eighth of the cost. Projected: −$46 a week, p95 unchanged.
Raise the identity source's concurrency window?
The provider throttle held 412 items this week for a median 21s. A wider window clears the queue at no added spend; the ceiling per run is unchanged.

nothing changes until a draft is committed · a promotion is logged with its actor and demotes itself on a breach

Spend by stepAsk →

Reach a verdict $54211%
Draft resolution $52010%
Implement $4208%
Correlate $3747%
Draft the report $2104%

security · stage by stageOpen the workflow →

Triage 11,200$336
Investigate 4,182$1,458
Report 1,240$445
Contain 312$150
Tune 8$151

detection-tuning is discovered and unpriced · price it

ExceptionsAsk →

Verdict lacks cited evidence security · security-investigations · step 4 · 14 this week, all recovered in line Escalating×14
Identity source throttled security · edr_api rate limit · items held with state intact, resumed on their own Watching×3
PR description fails the template repo-tasks · resolved by widening the check's context, 26 Jul Resolved×0
Lines 37 live $976 / $1,357 today yield 98.7% 1 throttle · 3 held
security-investigationsautonomous · monitored 9 live $248 / $334 98.9% open the line →
support-resolutionssupervised first runs 12 live $284 / $430 99.1% open the line →
repo-taskssupervised first runs 2 live $80 / $87 96.2% open the line →