Live Solana Risk Intelligence
RugBurn
Before a mint becomes a position, RugBurn asks what the chain already knows. It turns fragmented Solana token signals into evidence a trader, group, builder, or agent can act on.
90.4%
v3.2.2
90.9%
v3.3 shadow
177
samples
Role
Strategy, scoring, architecture, frontend, backend, API, bots, admin, launch.
Stack
Next.js, React, Go, Supabase, Better Auth, Birdeye, Helius, Telegram, Discord, MCP.
Status
Live product. Evidence-first scanner. Public claims kept honest while calibration matures.
Intelligence Doctrine
Evidence before confidence.
RugBurn does not hand users a naked score and ask for trust. It shows the pressure points: who can still change the token, where liquidity sits, how holders are distributed, whether the market is old enough to read, and when missing evidence should lower confidence.
01
Mint and freeze authority
02
Sellability and tax pressure
03
Liquidity depth and venue risk
04
Top holder concentration
05
Deployer exposure
06
Market stress and maturity
07
Metadata coverage
08
Hard score caps
The problem was speed.
Solana token decisions move faster than manual investigation. Traders jump between tools, moderators see suspicious mints before context is gathered, and agentic trading systems are starting to automate decisions without a reliable risk gate.
RugBurn answers the moment before action: paste the mint, expose the evidence, cap false confidence, and make the next move harder to fake.
The product had to stay honest.
New-token data is incomplete by nature. Provider credits are real. Accuracy can hide weak failed-token separation. Public language cannot pretend the system is financial advice or complete AML coverage.
The product language became stricter: risk evidence, workflow guardrails, confidence notes, and calibration truth.
System Flow
One evidence core. Many surfaces.
Dashboard, bots, API, MCP, and admin tooling stay useful because they draw from the same scan doctrine.
01
User, bot, API client, or MCP client submits a Solana token mint.
02
Authentication, tier, quota, and rate-limit checks run before expensive work starts.
03
The Go worker gathers token evidence from market and Solana data providers.
04
The scoring layer produces facets, hard caps, confidence notes, and final risk level.
05
Results persist to Supabase for dashboard reads, cached API reads, and calibration review.
06
Dashboard, Telegram, Discord, API, MCP, and admin surfaces read from the same evidence core.
Campaign Artifacts
Flyers built like field intelligence.
The launch banners carry the product mood: heat, warning, and restraint. They make RugBurn feel active without turning the scanner into spectacle.
Shipped Work
Built as a product, not a demo.
RugBurn includes the operating system around the scanner: subscriptions, quotas, admin controls, API access, bot workflows, calibration review, provider-cost discipline, and launch strategy.
Dashboard scanner with evidence-rich token reports
Wallet-risk surface
API v1 with quotas and scopes
MCP server for agentic clients
Telegram and Discord scan workflows
Admin subscription and usage controls
Queued email infrastructure
Birdeye-enriched scan path
Calibration Truth
The score is not allowed to lie.
A calibration snapshot showed strong overall accuracy, but weak failed-token separation. That changed the product posture: do not market perfect rug prediction. Market evidence, confidence, and pre-action guardrails.
90.4%
v3.2.2 accuracy
90.9%
v3.3 shadow accuracy
177
Calibration samples
16
Known failed tokens
My Role
End to end ownership.
Product strategy
Risk scoring doctrine
Backend architecture
Frontend implementation
API and MCP design
Admin operations