OMNI BOT →
SAMPLE DATA — not a live desk. Two of the block events below are a real, unedited capture from running the pipeline; the rest illustrate what open/closed trades look like. Set NEXT_PUBLIC_DESK_API_URL to your deployed backend to go live.

AUTONOMOUS RESEARCH DESK · SOLANA MEMECOINS

Seven agents. No people.
No hesitation, no revenge trades.

Watch it work. A mint lands, four agents open a file on it in parallel, a Chief of Staff scores the case, and the trader writes the exit before the entry. The same way, every time — that’s the whole point.

LIVE DISPATCH

THE FLOOR · SAMPLE ROUTE · CASE 1/1 · NEBCAT

Idle — waiting on the next pump.fun deploy

STAGE 2 · RESEARCH, IN PARALLEL

THE ROUTE

One mint, four stages, no discussion

CLICK ANY AGENT ABOVE TO OPEN ITS FILE

Chief of Staff never trades — it reads four independent briefs, scores the case against the rulebook, applies the SAFETY veto and the daily loss cap, and sizes it. SAFETY can kill any candidate at any score, and a kill does not get appealed. That is the whole reason a bad night stays a bad night instead of a blown account.

AGENTS

7

HUMANS IN LOOP

0

DECISION TIME

~11s

RULES VERSION

v3

MODE
PAPER
RULES
v3
OPEN POSITIONS
1
BLOCKED (24H)
2
UPTIME
14h 0m

THE TAPE

Every decision prints itself, in public

New rows land here the moment the route above finishes a case — including the ones the desk refused to trade.

SymbolMintStatusWhen
MOCK2715HfXg…Yjm9BLOCKED4m ago
MOCK47407GMq…6PCRBLOCKED11m ago
NEBCATcKaF…qjAVOPEN36m ago
NEBCATcKaF…qjAVSIZE37m ago
RUGWATCH6HsY…irDCCLOSED1h ago

THE BANKROLL

1 SOL to start · paper money only

Every position sized against real available balance — the desk refuses a new entry once the bankroll's fully committed, same as it would with real funds.

TOTAL EQUITY

0.998 SOL

TOTAL RETURN

-0.3%

AVAILABLE

0.901 SOL

COMMITTED

0.081 SOL

REALIZED P&L

-0.018 SOL

UNREALIZED P&L

+0.015 SOL

OPEN / CLOSED

1 / 1

WIN RATE

0%

BANKROLL COMMITTED8%

THE RULEBOOK

Rules v3 — rewritten every night, applied by hand

Every night the desk reads its own losing trades and proposes new weights and thresholds. The diff gets written to a file. It is never applied automatically.

SCORE WEIGHTS

HYPE30%
CHATTER25%
FORENSICS45%

THRESHOLDS

Min score to size65
Max top holder %15%
Max bundler %30%
Base size0.04 SOL
LAST NIGHTLY REVIEW · 5h agov3 PROPOSED

Win rate 33% is weak — raising the score bar from 62 to 65.

4/6 closes were stop-outs — cutting base size to 0.04 SOL.