v1v2
role
trust
Loopmason·Northwind Technologies/Whole organization

Command Center

Claude Code health across 500 engineers · last 30 days

Adoption
83%
+5%414 of 500 active
86 dormant — activation is the top lever
Sessions with substance
512
of 880 sessions today
Substance costs real work to fake — bare counts don't
Tokens used (MTD)
31.4B tokens
+9%of 37.5B tokens allowance
84% of allowance — plan fit, not waste
Craft distribution
74
+2avg · 51% of engineers ≥70
+17 moved into higher bands (30d) — movement, not a leaderboard
Recoverable now
5.5B tokens
18% of MTD tokens
POLICY 2.1B tokensCOACHING 3.5B tokens→ act in Token Efficiency
Open risk flags
15
5 critical
15 risky patterns flagged in-session

Activity

Sessions and active engineers · last 30 days

SessionsActive users

Real work

Inferred work threads across the org — the track layer between session and standup

Work Summaries
Infra 4Bug fix 7Feature 8Refactor 4
  • Saved-filters panel rebuild
    web-storefront · 4 sessions · 4 engineers · 17 cycles
    2.8h
    19M tokens
  • Token theming rollout
    design-system · 3 sessions · 3 engineers · 30 cycles
    5.8h
    12M tokens
  • Checkout retry logic hardening
    web-storefront · 3 sessions · 3 engineers · 15 cycles
    3.1h
    10M tokens
  • Usage-metering accuracy
    billing-core · 2 sessions · 2 engineers · 9 cycles
    2.2h
    10.0M tokens

Threads are inferred from session telemetry — clustering and titles are PROXY. Evidence of agent-assisted activity only: it does not verify delivery, and coverage of non-agent work is unknown by design.

Departments

Adoption & health by department

Compare teams
DepartmentEng.AdoptionCraftTokensRisks
Product Engineering
Priya Nair
182
86%
7410.3B tokens7
Platform
Marcus Feld
121
91%
818.4B tokens3
Infrastructure & SRE
Deepa Rao
88
72%
685.4B tokens2
Data & ML
Tom Weaver
109
79%
717.2B tokens3

Coaching throughput

What coaching delivered — throughput, not names

Mentoring
34
recs adopted (30d)
2.1B tokens
tokens recovered (30d)

This shows what coaching delivered — never who was coached — so it reads the same in both trust modes. At this altitude you steer with the craft distribution and this throughput, not names.

Executive summary

Auto-generated · every claim measured or labeled an estimate

Token detail

  • Adoption is 83% and rising (5% MoM). ~86 engineers are still dormant — activation is the single biggest lever left.

  • Average craft posture is 74/100 (+2 pts); 51% of active engineers sit at ≥70. Teams are improving most on plan-mode and verification discipline — using the agent well, not just using it.

  • 5.5B tokens/mo of tokens is optimizable (18%) — mostly model right-sizing and low cache-hit. Projections are audited: the ledger in Token Efficiency shows what each acted move actually saved, including the moves that saved nothing.

  • 27 risky patterns (secrets, destructive commands, prod access) were flagged in-session — measured directly from telemetry.

Risk flags

15 open

All alerts
  • low
    Prod access· 23h ago

    Session touched production database credentials

    Liam Patel
  • critical
    PII in transcript· 2d ago

    Customer email addresses appeared in transcript

    Aditi Turner
  • medium
    Prod access· 1d ago

    Session touched production database credentials

    Viktor Bose
  • critical
    PII in transcript· 1d ago

    Customer email addresses appeared in transcript

    Viktor Bose
  • low
    PII in transcript· 2d ago

    Customer email addresses appeared in transcript

    Blake Vidal