🧪 Token Launch Lab

An AI red team that tries to kill your token launch — before the market does.

Token launches are irreversible. One bad vesting schedule, one missing audit, one weak narrative, or one regulatory blind spot can torch a launch in public. Upload your whitepaper or TGE doc and four adversarial AI reviewers read it, then hand you back a GO / NO-GO decision, the exact things to fix first, and the evidence behind every finding.

▶ Try it live: token-launch-lab.onrender.com — free instance, first load may take ~40s to wake up. No key? Click View sample (HarborUSD).

---

💡 What it does

Most AI tools help you build launch materials. Token Launch Lab does the opposite — it attacks the launch you already have. Drop in a whitepaper, tokenomics doc, or TGE spec and it runs a pre-mortem: what will kill this launch, ranked, with the fix and the quote from your own document that proves it.

              YOUR WHITEPAPER / TGE DOC  (PDF · Markdown · text · paste)
                              │
        ┌─────────────┬───────┴───────┬──────────────────┐
        ▼             ▼               ▼                  ▼
   📉 DUMP RISK   🛡️ PROTOCOL RISK  ⚖️ REGULATORY RISK   🐦 NARRATIVE (CT)
        └─────────────┴───────┬───────┴──────────────────┘
                              ▼
                        📄 RISK REPORT
             GO / NO-GO score · Fix-first P0 list · evidence · remediation

---

⚔️ The four adversaries

Reviewer	Attacks
📉 Dump risk	Investor/team unlocks, TGE float, liquidity & market-maker allocation, launch-day sell pressure
🛡️ Protocol risk	Audit status, emergency pause authority, multisig & treasury control, launch readiness
⚖️ Regulatory risk	Sale/distribution mechanics, jurisdiction & eligibility exposure, security-likeness
🐦 Narrative risk	How Crypto Twitter ("CT") will react — points-farm optics, unsubstantiated claims, promise-vs-mechanics gaps

Every finding cites a near-verbatim quote from your document. If the document doesn't say it, the model doesn't claim it.

---

🚀 Quickstart

npm install

Add your GLM API key to a local .env file (copy from .env.example):

# .env
GLM_API_KEY=your-key-from-open.bigmodel.cn

Then start the app and open it in your browser:

npm run ui
# → http://127.0.0.1:3000

Upload a whitepaper / TGE doc (or paste the text) and hit Generate risk report. A full four-adversary report takes ~1–2 minutes. No key yet? Click View sample (HarborUSD) to see a demo report instantly.

The GLM key lives only in your server environment — end users never see or supply it. .env and generated reports/ are gitignored.

---

🧭 What's in a report

• GO / NO-GO decision with a 0–100 readiness score.
• Fix first — the P0 blockers you must clear before announcing, each with a concrete fix.
• Risk by adversary — a card per lens (Dump / Protocol / Regulatory / Narrative) with the worst severity and count.
• Full findings — every risk with severity, confidence, the evidence quote, why it matters, and remediation.
• Export PDF — a clean, printable report you can send to a co-founder or counsel.
• Share — a read-only link to the generated report.

---

🛡️ Safety boundaries

This is defensive launch-risk review only.

• ❌ Does not generate exploit instructions or attack payloads.
• ❌ Does not give legal advice — regulatory items are risks to review with qualified counsel.
• ❌ Does not produce harassment, misinformation, or market-manipulation content.
• ✅ Every finding is grounded in a quote from the document you provided.

---

🔧 Config

Variable	Default	Purpose
GLM_API_KEY	—	Required. Your GLM (Zhipu) API key.
GLM_MODEL	glm-4.6	The GLM model used for analysis.
GLM_BASE_URL	https://open.bigmodel.cn/api/paas/v4/chat/completions	GLM chat-completions endpoint.
PORT	3000	Local server port.

---

📂 Repo map

Path	What's inside
src/analyze.js	Reads the uploaded doc (PDF → text), calls GLM, shapes the risk report
src/demo-server.js	HTTP server: upload endpoint, report storage, share links, static UI
web/	The product UI (upload screen + report dashboard)
tge-spec.md · outputs/	The built-in HarborUSD sample report
.env.example	Copy to .env and add your key

---

Built for crypto founders who'd rather be killed in private than in public.

Contributor: Elena Xiao <leini8891@gmail.com>