J
JARVIS OS
Reflect Mirror
The Board Room

How Reflect Mirror is making decisions.

Every component, every model, every Claude session, every conversation between them — and how those communications turn into wise, researched, evidence-based decisions. This is the nervous system of the project.

Components Speaking
40
nodes in the graph
Active Pathways
29
communication edges
Consumer Calls
286
in the last 24 hours
Scheduled Runs
99
all clean
OPUSGRANTGPU FLEETTHE BOARD ROOMtap any node to learn what it does
← swipe the graph to pan →
JARVIS core
ML automation
Knowledge defense
Product / LLMs
Self-improvement / Fleet
Data sources
· tap a node for details · hover to highlight connections ·
How a decision happens

Every two minutes, the anomaly_engine reads the live training logs from all 5 GPU machines and flags anything pathological. When it finds something worth killing, the kill_rule_engine picks it up, asks Qwen3 235B Instruct to design a replacement experiment, and drafts the kill+launch plan. That plan lands in your queue — JARVIS never executes destructive operations. You approve from Telegram or the dashboard, and the action moves to Opus or you to run.

Every step is calibration-logged. Every model choice is anchored by a discipline rule (235B-class default, smaller models earn slots only via per-task calibration). Every prompt is versioned and A/B tested before promotion. The whole system is built so that every decision is wise, researched, and evidence-based.

Recent Decision Chains

anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
brain_cycle_digest ran
Complete
🧠
JARVIS
scheduler tick
🧠
brain_cycle_digest
manifest_validator ran
Complete
🧠
JARVIS
scheduler tick
🛡
manifest_validator
brand_compliance ran
Complete
🧠
JARVIS
scheduler tick
💬
brand_compliance
anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
anomaly_engine ran
Complete
🧠
JARVIS
scheduler tick
⚠️
anomaly_engine
Unified status check
Complete
⚠️
anomaly_engine
rule scan
🛡
manifest_validator
manifest gate
📚
memory_auditor
hygiene scan
💬
brand_compliance
voice scan
🧠
JARVIS
merged: 2 anomalies, 0 kills
Unified status check
Complete
⚠️
anomaly_engine
rule scan
🛡
manifest_validator
manifest gate
📚
memory_auditor
hygiene scan
💬
brand_compliance
voice scan
🧠
JARVIS
merged: 2 anomalies, 0 kills
Unified status check
Complete
⚠️
anomaly_engine
rule scan
🛡
manifest_validator
manifest gate
📚
memory_auditor
hygiene scan
💬
brand_compliance
voice scan
🧠
JARVIS
merged: 2 anomalies, 0 kills

Local LLM Portfolio

Qwen3 235B Instruct
primary text
13 calls in 24h
Qwen3-VL 235B
primary VLM
1 call in 24h
Llama 3.3 70B
calibration
9 calls in 24h
DeepSeek-R1 32B
calibration
0 calls in 24h
Nomic Embed
embeddings
0 calls in 24h
Chat with JARVIS
Connecting...
Telegram
💬
No messages yet
Just talk to JARVIS like a person. No commands needed — ask anything about the project.
Same conversation as your Telegram chat. JARVIS thinks before replying.