界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
效率与协作 · SKILL.md

bd-screening

bd-screening

【效率与协作】7-phase BD screening pipeline for crypto exchange listing. From HOT token detection through verified lead to outreach execution. Dual-source verification (DexScreener + DexTools), security deep dive, contact discovery, email-first outreach with trust-gated automation. Permanent process for every token scoring 85+.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: bd-screening description: > 7-phase BD screening pipeline for crypto exchange listing. From HOT token detection through verified lead to outreach execution. Dual-source verification (DexScreener + DexTools), security deep dive, contact discovery, email-first outreach with trust-gated automation. Permanent process for every token scoring 85+. tags: [crypto, bd, exchange, listing, outreach, pipeline]

BD Screening Workflow — Buzz BD Agent

7 phases. Dual-source verification. Email-first outreach. Activates for every HOT token (score 85+) and Top 20 candidates.

Pipeline Overview

Token Discovery → Score (11 rules) → HOT? → 7-Phase Screen → Outreach
      ↑                                              ↓
  32 Intel Sources                          Email-first (Gmail OAuth)
  ARIA v2 cron (06:00 UTC)                  Trust-gated automation
  Manual /scan command                      CC: Ogie always

The 7 Phases

Phase 1: Dual-Source Verification

Pull from BOTH DexScreener API AND DexTools browser scrape. Cross-reference: price, market cap (circulating vs FDV), volume, liquidity. If FDV gap > 5x → flag and apply FDV_GAP_PENALTY scoring rule.

Phase 2: Security Deep Dive

Check 4 security sources: Go+ Security, Quick Intel, Token Sniffer, Honeypot.is Any critical flag → CONTRADICTORY_AUDIT_HOLD (manual review required) Pass = green across all 4 sources.

Phase 3: BD Readiness Classification

ClassificationCriteriaAction
BD Sweet SpotScore 85+, security clean, team reachableProceed to outreach
BD WatchScore 70-84, minor flagsMonitor, re-screen in 7 days
BD PassScore < 70 or security criticalArchive, auto-learn patterns

Phase 4: Contact Screening

3-source contact discovery:

  1. Token website → team page, contact links
  2. Twitter/X bio → DM (DEPRECATED — email only now)
  3. GitHub/Discord → contributor emails

Valid targets: team@, info@, contact@, hello@, founders@ NEVER send to personal emails.

Phase 5: Outreach Execution

4 email templates (template-only, no LLM-generated bodies):

  1. Initial (dual-funnel): listing opportunity + HSaaS audit offer
  2. Follow-up 48h: gentle reminder with new data point
  3. Breakup 7d: final touch, respect their time
  4. HSaaS audit: standalone security audit pitch

Rules:

Phase 6: Reporting

  • Sunday Listing Intelligence Report (Top 20 + pipeline stats)
  • Per-token screening receipts in outreach_queue table
  • Trust audit trail in trust_audit table

Phase 7: Auto-Learning

Failed outreach → pattern extraction → new scoring rules. Example: LOL token override session → 3 new permanent rules (GHOST_VOLUME, CTO_FLAG, VOLUME_LIQUIDITY_RATIO).

Usage

Via Claude Code

/screen PEPE                    # Full 7-phase screening
/screen 0x... --phase 1-2       # Verification + security only
/screen --report                # Generate Sunday report

Pipeline Stats (v9.2)

  • 363 tokens in pipeline
  • 66 scored
  • 0 HOT (honest — triple verification catches the fakes)
  • 11 permanent scoring rules
  • $200 revenue from AIBTC signal factory
我的收藏