RLL · AGENT TEAM

Reaves Labs and Learning · 210 autonomous agents · 4 governance tiers

🧠 3D Company Brain

Live 3D visualization · 210 agents · 184 parent-child connections · click any node for full D44 v2 job description · NULL seats glow red

Direct JSON · full agent index · same shape as brain-http /v1/agents-bulk · public read-only mirror

Last sync timestamp · agent count · scrubbed fields list · mirror diagnostics

cloud surface · synced from mac1 every 5min via com.rll.brain-supabase-mirror

source of truth: ~/RLL/state/agents/skills/*.json on mac1

doctrine: D30 — 0% cloud for actions; this is a read-only viewer mirror

trust rule: revenue fields scrubbed via regex patterns · audit trail in meta.json.scrubbed_keys_sample

rollback: agents-bulk.last-good.json holds previous validated snapshot

SCOPE — what the cloud surface deliberately does NOT do

Reports tab (agent execution traces): stays local on mac1 — operational logs may contain sensitive data per D30 trust boundary

Brain queries (/v1/ask): stays local on mac1 — proxying queries through Netlify would expose brain-http auth + leak query terms

Task execution / model calls / ledger writes: stay local on mac1 — D30 §1 binding constraint (0% cloud for actions)

For Reports + brain queries, open AGENT TEAM PWA on mac1 directly: open ~/Desktop/AGENT\ TEAM.app