What the blocks mean
Structured evidence on a review is separate from the editorial article. Labels map to on-page anchors:
- Risk Passport (
#risk-passport) — a compact due-diligence card: passport status, open-model score, coverage, watchouts, and unknowns. Not a price forecast. - Sources (
#evidence-ledger) — the evidence ledger: each row is a fact, its value, how strongly we trust it, and where it came from. - Checks (
#hands-on) — desk checks against public pages and repos; hands-on device/account tests stay Not run until we actually perform them. - Incidents (
#incidents) — curated incident notes with sources. A starting checklist, not a complete history. Missing entries are not proof of safety. - Research (
#research-notes) — short automated notes from public GitHub releases for the linked repository. Useful signal, not a rewrite of the review. - What’s new (
#whats-new) — bounded updates when tracked evidence moves. Only the marked update section is replaced; the rest of the article is not silently rewritten.
Fact status meanings
Every Sources row carries a verification status:
- Verified — cross-checked, or backed by a primary/trusted source we accept for that field.
- Observed — seen in a source, but not yet at our verified bar. Volatile market figures often stay observed until a second source exists.
- Conflict — sources disagree. Treat with extra caution until resolved.
- Stale — older than our freshness window for that field type. May still be directionally useful.
Passport status meanings
- Verified — enough tracked evidence supports a full passport reading.
- Partially verified — useful evidence exists, but material gaps remain.
- Insufficient evidence — coverage is below our model threshold. We do not invent a comforting mid-range score to fill empty space.
The risk score is 0–100 from TokenScouts’ open, deterministic model. Higher usually means stronger evidence-backed standing in that model — not a recommendation to use or invest.
How scoring works (open model)
Scores are computed by a versioned rules file per entity type (coin, wallet, exchange), not by an LLM choosing a number. Same inputs produce the same outputs. Model files and engine live with the TokenScouts evidence pipeline; the public methodology target is to mirror those rules on GitHub as they mature.
Writers may explain verified facts. Automated critics block unverified claims and investment hype such as guaranteed returns.
Not financial advice
TokenScouts is due-diligence research for ordinary readers. We do not promise safety, returns, or regulatory outcomes. Always verify primary sources linked on each page and follow the laws that apply to you.
What auto-update does — and does not
Does: refresh API observations into the evidence store; re-check conflicts; recompute open-model scores; refresh on-page evidence blocks from caches; optionally update only the marked “What’s new” section when material signals change; create WordPress drafts for human review.
Does not: auto-publish live pages without an explicit decision; invent hands-on test results; silently rewrite an entire review as “AI refreshed everything”; claim a complete incidents catalog.
Hands-on honesty
- Desk — checked from public documentation or URLs without creating an account.
- Hands-on — requires a real product run (device, signup, deposit, withdrawal, and similar). We leave these Not run until TokenScouts records a real result. We never invent them.
Corrections
When a fact or score was wrong, prefer a dated note in What’s new / Sources over a silent overwrite. Methodology and model versions are recorded with the evidence pipeline.
Page draft version: methodology-page-v1. Editorial source of truth is maintained with the TokenScouts site operations docs; live publication is a human gate.