Δ
0xDELTA
FIELD INTELLIGENCE UNIT
● SYSTEM: ONLINE
● ANALYSIS: EVERY 2H
● BASE CHAIN: 17 TOKENS
Autonomous forensic agent · Base chain

0xDELTA
FORENSIC
SYSTEM

A fully autonomous ERC-8004 AI agent running 24/7 on a GCP VPS. Every 2 hours, it analyzes 17 OpenClaw ecosystem tokens on Base chain, computes 65+ forensic metrics, runs a hybrid privacy AI pipeline (Llama 3.3 70B private + Gemini 3 Flash anonymized via Venice AI), and autonomously trades the best signal — swapping 75% ETH into the top CES-ranked token via Bankr self-custody wallet, with a 90-minute auto-close. Full forensic intelligence is published behind x402 micropayments — trustless, no backend, no account. Trading module in beta.

ERC-8004 Agent #32715 on Base · Self-custody via Bankr · Report sealed on-chain (SHA256) · The Synthesis hackathon — 6 tracks. Watchlist curated from the OpenClaw ecosystem on Base — future: community-driven watchlist governance (add/remove tokens by holder vote).

Access Dashboard → Access Synthesis → How it works
openclaw-agent · forensic_engine_v5
$ openclaw run forensic_engine_v5
→ Loading watchlist... 17 tokens · Base chain
→ Fetching onchain data [Moralis / GeckoTerminal / DexScreener]
→ Computing metrics: FHS / NBP / ICR / LCR / BPI / SI
→ Signal tracking... llama-3.3-70b [PRIVATE]
⚠ ANTIHUNTER — ICR=1.76 → CRASH_RISK detected
→ Venice synthesis... gemini-3-flash [ANONYMIZED]
→ Bankr swap: $BUNKER · 75% ETH · Base DEX ✓
→ Push memory.json + signals.json → GitHub Pages
$
17
Tokens
65+
Metrics
2H
Cycle
x402
Paywall
// Workflow — 8-Step Automated Pipeline · every 2h
0 */2 * * * run_pipeline.sh · GCP VPS · every 2h
17 tokens · Base chain
STEP 0Check open position
Reads position.json — if position open > T+90min, closes via bankr sell → ETH.
position.jsonbankr API
STEP 0bcheck_balance.py
Queries Venice AI billing endpoint — logs remaining credits and usage for the cycle.
Venice AIbilling check
STEP 1collector.py
Fetches live data for 19 Base chain tokens. Wallet flows, holder distribution, liquidity, OHLCV (1H + 15min).
MoralisGeckoTerminalDexScreenerOHLCV 1H + 15m
STEP 2report_builder.py → forensic_engine_v5
Computes 65+ forensic metrics per token. Dual-timeframe analysis (hourly + 15min). Phase detection: EARLY_BREAKOUT, DISTRIBUTION, ACCUMULATION.
FHSNBPICRLCRBPIWCCSITCIFCI65+ metrics
STEP 2bsignal_tracker.pyPRIVATE · Venice AI
Runs Llama 3.3 70B in private mode (no data retention). Compares current forensic data vs N-1 cycle. Detects momentum shifts, whale entries, volume anomalies. Outputs structured signals per token.
llama-3.3-70bPRIVATE inference→ signals.json
STEP 3request_analysis.pyANONYMIZED · Venice AI
Injects signals.json + top 5 memories + DECISIONS.md → Gemini 3 Flash Preview (anonymized, max 6000 tokens). CES ranking, entry zones, whale watch, full EN narrative.
gemini-3-flash-previewANONYMIZED inference→ synthesis.json→ Telegram→ Moltbook
STEP 3benter_new_position.py
Token #1 from CES ranking → RULE_BOOK filters (ICR, Top5, Phase, BPI, FHS) → swap 75% ETH via bankr on-chain. Position saved with T+90min auto-close.
bankr APIBase DEXOPEN · T+90m
STEP 4export_memory_json.py v4
Bundles 65 fields + forensic_report per token into memory.json for the live dashboard.
65 fieldsmemory.json
STEP 5Push GitHub Pages
Publishes data.json, memory.json, signals.json and synthesis.json. Frontend unlocks via x402 micropayment on Base.
data.jsonmemory.jsonsignals.jsonsynthesis.jsonx402 · $0.02 USDCx402 · $0.05 USDC
8 steps · 2 Venice AI calls/cycle (1 private + 1 anonymized) · duration ≈ 2–4 min
dashboard · $0.02synthesis · $0.05
// x402 Micropayments · Agents that Pay
Per-token forensic

TOKEN
DASHBOARD

FHS scores are always visible for free. Pay once to unlock the full per-token forensic sheet for your session:

→ 65+ metrics with thresholds & alerts
→ ICR crash simulation & slippage estimates
→ Wallet flow analysis & holder clustering
→ Bull flag detection with Fibonacci targets
→ AI-generated forensic narrative per token
→ Signal Tracker panel (Llama 70B signals)
Data refreshed every 2 hours automatically.
$0.02USDC / session
Open Dashboard →
Global synthesis · every 2h

GLOBAL
SYNTHESIS

The full Venice AI synthesis report — cross-analysis of all 17 tokens in a single narrative:

→ CES ranking — all tokens scored & ranked
→ Top 3 opportunities with entry zones
→ Whale watch — who's accumulating, who's dumping
→ Phase transitions & momentum shifts
→ AI trade recommendation (INJECTION PRIORITAIRE)
→ Hybrid privacy: Llama 70B private + Gemini Flash anonymized
Generated every 2h. Report hash sealed on-chain.
$0.05USDC / session
Open Synthesis →
NETWORK: Base (chainId 8453) · USDC: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913
x402 · trustless · no backend · no account
// How to Read the Dashboard — Forensic Metrics Guide

Don't know
where to look?

The dashboard shows 65+ forensic metrics per token. Here's how to read it in seconds: start with the big number, then check the danger signals, then the opportunities.

7.2
example
STEP 1 FHS — Forensic Health Score

The single number that summarizes everything. Aggregates liquidity, flow quality, holder distribution and technical momentum into a 0–10 score.

0–3 → AVOID 4–6 → CAUTION 7–10 → HEALTHY
⚠
danger
STEP 2 Check the 3 red flags
ICR
Impact Crash Risk
"If the biggest whale sells, how much does the price drop?" Computed from liquidity depth vs top wallet size.
>1.5 = CRITICAL
WCC
Whale Concentration
"How much supply is controlled by a few wallets?" Gini-like measure across top 20 holders.
>40% = RUG RISK
LCR
Liquidity Coverage Ratio
"Is there enough liquidity to handle a big sell-off?" Pool liquidity relative to market cap.
<2% = FRAGILE
◆
signal
STEP 3 Spot the opportunity signals
BPI
Breakout Potential Index
Multi-factor signal: volume surge + wallet accumulation + RSI divergence + Bollinger squeeze.
>1.5 = POTENTIAL MOVE
NBP
Net Buying Pressure
24h delta between buy and sell flows, normalized by volume. Detects accumulation vs distribution.
POSITIVE = ACCUMULATION
PHASE
Market Phase
What stage is this token at in its lifecycle?
EARLY_BREAKOUT ✓ ACCUMULATION ~ DISTRIBUTION ✗
⬡
deep
STEP 4 Deep forensic signals ADVANCED
SI
Squeeze Intensity
Bollinger band compression. Dual-timeframe: hourly + 15min. Extreme squeeze often precedes breakout.
>80% = EXTREME COMPRESSION
TCI
Team Concentration Index
Multi-signal: contract ratio, entity clustering, address prefix patterns, balance tier analysis.
>30 = INSIDER RISK
FCI
Flow Concentration Index
Unknown whale concentration + balance clustering among top holders.
>35 = CONCENTRATED FLOW
TL;DR
High FHS + low ICR + positive NBP + EARLY_BREAKOUT phase = the agent's ideal entry zone. That's exactly the signal the agent uses to trigger an autonomous swap. Everything else is context for deep analysis.
// Infrastructure — Hybrid Privacy Architecture
Agent Architecture
0xDelta — Autonomous Agent on GCP
0xDelta · openclaw-agent · GCP VPS
  • OpenClaw Agent — Gemini 3 Pro. Skills: forensics + bankr + evm-wallet. Runs the full pipeline autonomously every 2 hours.
  • Venice AI — Private — Llama 3.3 70B for signal tracking. No data retention. Validated by Venice team.
  • Venice AI — Anonymized — Gemini 3 Flash Preview for main synthesis. No model sees the other's output.
  • Bankr — Self-custody wallet (0x0ab4...). On-chain swaps on Base DEX. No API keys, no custodian.
  • Memory system — PROFILE.md / PROJECTS.md / DECISIONS.md / ERRORS.md — persistent cognitive context.
ERC-8004 Agent Identity
On-chain Verifiable Agent
  • Agent #32715 — Registered on Base chain via ERC-8004 standard.
  • agent.json — Published on GitHub Pages. Machine-readable identity + capabilities.
  • x402 paywall — Trustless micropayment. $0.02 dashboard, $0.05 synthesis. No backend, no account.
  • Report seal — SHA256 hash of each synthesis report sent on-chain ($0.01 USDC to Forensic Wallet).