Words on this page
| Word | Meaning |
|---|---|
| CA, mint | A Solana token's contract address: the one identifier no other coin can share. |
| Check tech | The button that asks Golem to test a coin's website now. |
| Receipt | Golem's public record of one check: what it saw, what it could not test, its verdict. |
| Verdict | One rung of five, from "insufficient public evidence" to "public product observed with integration evidence". |
| Tech works | The gold badge: the top rung, on the coin's own website. Stands for 7 days. |
| Candidate | A coin Golem is watching, with a state and a rank. |
| golem-be, golem-fe | The backend that does the work, and this app. |
What the badge does not claim
Tech works means one bounded, public check passed: the declared website rendered, with working controls and live data from its own servers or from Solana. It does not review the token's code, its contracts, its team or its claims, and no authenticated, paid or signed action is ever performed.
Read a receipt yourself
Every receipt is public: golem-api's token reads need no key. The app reads the same route through its own gateway, and the verdict is one field.
# Public token reads need no key. Locally: cargo run -p golem-bin (port 8080).GOLEM_API="http://127.0.0.1:8080"MINT="<mint address>"
curl -s "$GOLEM_API/v1/tokens/$MINT/technical-underwriting" \ | jq '{verdict: .receipt.verdict, confidence: .receipt.evidence_confidence}'Routes and where they show
| golem-be route | What it answers | In the app |
|---|---|---|
GET /v1/candidates | The ranked feed, 25 a page (max 100), cursor-paged | Explore → Golem research; row tech status |
GET /v1/tokens/{mint}/technical-underwriting | The technical receipt and its verdict | Coin page tech report, every Tech works badge, Scan, the hero check |
GET /v1/tokens/{mint}/dossier | On-chain, market, pool, web and social evidence | Coin page chain checks, Websites inspector, Scan |
GET /v1/tokens/{mint}/synthesis | Advisory AI synthesis (OpenRouter) | Coin page AI notes, fenced and never scored |
GET /v1/tokens/{mint}/assessment | The deterministic assessment | Coin page → Full research record |
POST /v1/me/token-analyses (+ /{id}, /{id}/stream) | Queue a check, follow it | Scan, the hero check, research record |
GET /v1/assessments/{id} | One recorded assessment | Scan, to confirm a finished check |
/v1/me/watchlists (+ items) | The viewer's watchlists | Explore → Watchlist, every Watch button |
/v1/me/entitlement, /v1/tokens/{mint}, …/timeline, …/state-history, …/evidence, …/social-links, …/social-velocity, /backtest, /replay | Forwarded by the gateway | Not read by any screen yet |
/v1/me/wallets, account/*, allocation-profile, alerts, strategies/*, orders, POST /v1/strategy/preview | Accounts, strategies, orders and alerts | Not in the app |
Four gaps between the app and golem-be main. Scan sendsfresh_request_idwith a check, which main'sPOST /v1/me/token-analysesrefuses (it accepts{mint}only). The website routes (/v1/website-candidates,/v1/me/website-discovery,/v1/me/website-renders) are missing too; both fixes are on golem-be'sfeat/scan-research-events. The agent chat routes behind Scan → Agent are on no golem-be branch. And/v1/candidateshas no tech verdict per row, so a list shows Tech works only for receipts already read one coin at a time.
Was this page helpful?