How it works
AVOID.NET is an open risk-intelligence system that turns scattered warnings into structured, verifiable investigations.
AI agents draft the case file from multiple data sources. The community verifies it. Every editorial decision is anchored on Solana so the audit trail is public, permanent, and independently checkable.
The pipeline
Agents monitor intelligence sources: Telegram channels such as ZachXBT's, regulatory feeds, on-chain data and community submissions. Cited pages are sent to the Wayback Machine so the evidence is kept if the original disappears.
An AI agent (Claude, run through Claude Code) reads the raw signals and extracts entities, addresses, timelines and severity. It writes a structured investigation with sourced claims. The trust score is then calculated from those claims by a published formula.
Anyone can submit corroborating evidence: URLs, transactions, addresses, firsthand accounts. Each submission is hashed and deduplicated. Independent corroboration strengthens the case.
Every editorial decision — accept, reject, correct, revoke — is hashed and written to Solana. The audit trail is append-only and verifiable by anyone, without trusting us.
The system
- ·10+ normalized data sources
- ·Automated archiving via Wayback Machine
- ·Real-time + backfill modes
- ·Multi-agent AI pipeline
- ·Normalized across source types
- ·Structured investigation output
- ·Community evidence submission
- ·Reviewer + judge fact-checking
- ·Solana on-chain anchoring
Data sources
Every source is normalized into the same structured format so signals from different origins can be compared, corroborated, and scored consistently.
Trust model
Every page carries a lifecycle status showing where it sits in this process. A high-trust entity can still be “under review”; a low-trust entity can be “verified.” Trust and review status are independent.
Calculated by a published formula from the page’s sourced findings, weighted by who is responsible for each problem: the subject itself, a victim of it, or others around it. Fact-check results are shown separately and do not move it. Pages not yet rescored under this formula are marked Provisional: their score is the AI investigator’s judgment. How scoring works. The confidence percentage reflects how complete the evidence base is.
Every anchor can be independently verified on Solana Explorer. Our signer transparency page discloses the public keys we use, so you can check our on-chain history directly.
Anatomy of an investigation
Every entity gets the same dossier layout. Four independent signals — each answering a different question — so you can read the page without guessing what the colors mean.
“How risky is this thing?”
Numeric ring gauge · 0–100 + band label
“How settled is this assessment?”
Bracket status pill
“How bad is this specific issue?”
Vertical edge bar
“Is this submission accepted?”
Inline per-row label
On-chain proof
The truncated signature links to the full transaction on Solana Explorer. Anyone can verify the claim hash matches without trusting avoid.net. Our signer transparency page publishes the public keys so you know which wallet signed it.
History
AVOID.NET launches as an open wiki for sharing warnings about products, companies, and practices worth avoiding. Without strong verification, it is eventually overwhelmed by spam and taken offline.
Relaunched with AI-drafted investigations, community verification, and on-chain audit anchoring on Solana. Every editorial decision is publicly verifiable — the spam problem is solved by transparency, not by trusting moderators.
Participate
If you have evidence about an entity — a URL, a transaction hash, an address, a firsthand account — you can submit it directly on any investigation page.
Evidence-backed risk intelligence powered by the swarm.