📖 How to drive this department in 30 seconds
- 🎯 Dispatch a task in Quick Dispatch — the engineer squad routes it to the right lane.
- 🔄 Watch Mission Lifecycle below — each glowing node lights when your task hits that stage.
- 🏆 Score climbs in Leaderboards; squad members trash-talk in the next panel 🔥.
- 🗺️ Open Roadmap → to see how each milestone fed the production-readiness score.
- ✅ Hover any KPI for the variable behind the number.
- ✅ Click any Lifecycle node to jump to its panel.
- ✅ Toggle 🎙 Monitor in Trash Talk for live audio jabs.
🔄 Mission Lifecycle idle
Every engineering mission flows through these 5 stages. The active stage glows.
Click a stage to inspect its tunable variables and jump to the relevant panel.
Filed
—
Routed
—
Applied
—
Verified
—
Handled
—
—
Plans / 24h
—
Tools / 24h
—
Agents / 24h
—
Missions / 24h
—
Verify pass
—
Verify fail
—
Streak
—
Inbox depth
—
Handled
—
Catalog tools
⚡ Quick Dispatch
🏆 Top 3 right now
Persona podium — the three engineers with the highest score this 24h.
Hover for rivals/allies. Click → Evolution HUD for the full picture.
👷 Engineering Squad — Historical Roster
Nine named engineers from antiquity through 2026. Hero of Alexandria leads;
the rest were chosen for what they actually are — chaos engine, product
absolutist, line doctrinaire, patent warrior, visionary, cursed artifact,
Skunk Works disciplinarian, hardware-shipper. Cards glow gold/silver/bronze
for the top 3 by leaderboard score. Hover any card for rivals + allies +
voice. Tap a card to open trash talk for that codename.
🔌 Live agent roster (raw lane status)
🌐 Browser Router
RAM- and task-aware browser picker. Replaces the static
selection_rules table on browser_squad. Picks recorded to
state/agents/browser-router-feedback.jsonl for the
outcome learner.
🧪 Smoke Team
smoke_router_v1 picks the best harness from
run_test_suite.py
per surface change (auth → cross_site_isolation; ui → mobile_tester;
voice → voice_pipeline; etc.). Records every run to the feedback
ledger so the learner can ratchet weights.
The Staging Suite iframe was promoted to its own
Staging Suite panel (PR1 2026-04-30) so it's no longer duplicated here.
📦 Software Catalog (159 sources / 427 tools)
🤖 Agent Admin
🛠️ Tools Admin
🎯 Mission Console
🧬 Self-Evolution Loop
Producer effectiveness over 24h with hourly trend curves, breaker
history, cost burn per slug, policy ratchet timeline, and a
“next-ratchet” preview from
mission_outcome_learner_v1.
📈 Producers — 24h trend
loading…
🛑 Breaker history (last 50)
no breaker events
💸 Cost burn (today)
no slugs over threshold
⏬ Next-ratchet preview
no producers near threshold
🔀 Handler-policy matrix
no policies recorded
📜 Policy-change timeline (last 100)
no ratchets recorded
🔄 Feedback Loop
Verify pass/fail trend, last 30 verifies with mission ids,
postmortem buckets with example incidents, top failure clusters,
learner-decision history, and squad confidence drift.
📉 Verify pass / fail (24h, 12 buckets)
no verifies in window
🧾 Recent verifies (last 30)
no recent verifies
🎯 Top failure clusters
no failure data
📑 Postmortem buckets
no postmortems yet
📊 Squad confidence trend
no squad data
🧠 Learner decision timeline (last 20)
no learner runs recorded
🏆 Leaderboards (last 24h)
🔥 Squad Trash Talk
0 jabs
📖 What is this?
- Squad members trash-talk each other when they pick winners or close findings.
- Lines stream in from
state/agents/engineering-competition.jsonl. - Toggle 🎙 Monitor to live-poll every 10s and pulse new chips into the panel.
- Each line is voiced by the codename's persona voice (Hero=onyx, Musk=verse, Jobs=echo, Tesla=nova, Edison=sage, Ford=ash, von Braun=ballad, Kelly=coral, Palmer=fable).
📡 HUD feed (auto-updates every 60s)
💬 Engineering Chat
Talks directly to
engineering_dispatcher_v1 which
fans out across the lanes. Each routed task is paired with
live catalog matches so reuse beats build.
Hey — describe a task and I'll route it to the right lane. Try "review the login flow" or "wrap pip-audit as an agent".