facts before scores
Version 0.1 publishes components, not a composite trust score: accountability level, contract-control level and factual flags, each with evidence and a date. A weighted score comes only after public consultation, once 30 or more projects are live.
Accountability · A0–A4
How checkable the operator is, not where it is based. A pseudonymous team is not a red flag in itself; hiding who runs a project is.
A Verified Tag needs A3 or higher, checked within the last 12 months. Expired checks fall back to A2.
Contract control · C0–C4
Who can change the code, read from the chain: proxy admin slots, Safe owners and thresholds, timelock delays, governors.
classification
14 primary sectors (exactly one)
5 audience positions (exactly one)
Positions shown with a dashed outline and "?" are suggestions from research, not yet confirmed by the project team.
flag catalogue
| Code | Severity | Statement template |
|---|---|---|
| IMPOSTOR_TOKEN | critical | A token using this project's name is trading at {address}. The project has not announced a token. |
| SANCTIONED_MATCH | critical | {subject} matches the {list} sanctions list. |
| CONTROL_SINGLE_KEY | caution | Contracts can be upgraded or paused by a single key ({address}). |
| SOURCE_UNVERIFIED | caution | Contract source code is not verified on {explorer}. |
| SITE_OFFLINE_72H | caution | Website unreachable for more than 72 hours (last checked {date}). |
| NO_ACTIVITY_90D | info | No on-chain activity in 90 days. |
| NO_LICENCE_DISCLOSED | caution | No {licence} licence disclosed for this {sector} application. |
| OPERATOR_UNDISCLOSED | info | No operator disclosed (A0). |
| DAO_INSIDER_MAJORITY | caution | Top {n} wallets hold {x}% of voting power. |
| DAO_MULTISIG_OVERRIDE | caution | A multisig can execute changes without a DAO vote. |
| AUDIT_PUBLISHED | ok | Audit published by {firm} ({date}). |
Flags state facts with a source and a date. DogeCompass never labels a project a scam, a rug or safe.
sector minimums
| Sector / audience | Expected | Flag if missing |
|---|---|---|
| Casino & GambleFi | A4: gaming licence on the regulator's register; 18+ and country limits | No gaming licence disclosed |
| Prediction Markets & SportsFi | A3+, plus regulatory status and country limits | Regulatory status not disclosed |
| Payments & Commerce | A3+, plus payment or VASP registration where required | No payment registration disclosed |
| RWA & Enterprise | A3+, and the issuer of the underlying asset named | Issuer not verified |
| Enterprise / Institutional audience | A3+ | Position shown as self-declared |
| Memecoins & Launchpads | Any level; contract control always shown | Level shown prominently |
Agentic daily sweep
Once a day an agent reads public sources (the official DogeOS list, the DogeOS blog, project sites, terms and imprints) using Firecrawl, and Claude extracts facts into a strict schema. The agent only suggests changes, each with a snapshot of its source. Editors approve or reject every suggestion; nothing the agent reads can publish a fact or a flag on its own.
Daily news
Every morning an agent searches Google News (through SerpAPI) for DogeOS, Dogecoin topics and every listed project. Known press-release wires and headlines about price predictions, presales or trading calls are dropped without a model call. Claude then screens the remaining headlines, keeps reporting and official announcements that matter to the ecosystem, reads each article with Firecrawl and writes a short brief in its own words with a verified quote from the article. Briefs publish automatically and always link to the original. Editors can correct or remove any brief, and every change is logged. We do not show publishers' photos: each story gets a generated pixel stamp instead.
Right of reply & changes
Every project can dispute any value: we acknowledge within 72 hours, resolve within 10 business days, and log the outcome. Before any flag that could harm a project's reputation is published (critical flags and flags about licences, sanctions or control), the project is notified and gets 48 hours to respond. Active impostor or phishing alerts are the exception: they go live immediately to protect users. Methodology changes are versioned and announced 14 days before they apply. Sponsors and funders have no influence on levels or flags.