Methodology

What counts as building

HEY records a ShipEvent when a project does something concrete: a release, a deployment, an integration, an SDK, documentation, or — for meme projects — a game, tool, bot, art drop or community event. Creative and community output is real output. A meme is never required to pretend it ships software.

Announcements, hype, giveaways and marketing posts are recorded but never counted as building. An event also has to be backed by a public source or better before it can affect a project's status, so a project cannot talk its way into looking active.

Commits are read too, a week at a time, and HEY reads what each commit changed (classifier commit-substance-v4). A week in which every commit only changed documentation or the README, dependency lockfiles, dependency or version updates (a manifest such as package.json, Cargo.toml or go.mod whose changed lines are all versions), generated files, images and other assets, data files (JSON, CSV and the like outside configuration), whitespace, or only CI workflow files is shown on the project's timeline as “documentation or maintenance only” and is not counted as building. Commits from bot accounts and from automated streams — one templated message repeated on a clock, or one carrying its own changing timestamp or figures — are never counted, and a week of nothing else is shown as automated. A busy week of 25 or more human commits counts only when at least 3 of a sample of 10 of its commits change code; otherwise it reads “mostly maintenance or automation”. Any other week needs one commit that changed source code, tests or configuration — a manifest change that is more than versions, or one whose patch GitHub did not return, is configuration. A week, or a sample, HEY has not read yet counts as it always has: an unread week never demotes a project. Merge commits are not counted. A merged pull request on its own is not a ship.

A repository's releases of one UTC day count once (since hbm-v23): when a team cuts many releases in a day, the newest stands for the day in activity status, Build Momentum and every count of building. Releases from different repositories on the same day count apart, and every release stays on the timeline with its source.

Contracts a project's deployer puts up each within 60 seconds of the one before count once (since hbm-v28; the same second from hbm-v24): a deployment script that creates several contracts in a run is one deployment in activity status, Build Momentum and every count of building. Deployments more than 60 seconds apart count apart, and every contract stays on the timeline with its creating transaction.

Activity status

Shipping

A meaningful, source-backed update in the last 7 days.

Active

A meaningful update in the last 30 days, or at least 2 in the last 45 days.

Quiet

No meaningful update for 30–60 days, and fewer than 2 in the last 45 days — a project that kept up that pace still reads Active.

Dormant

No meaningful update observed for over 60 days, or none ever, and only where HEY has a source it could have seen one through. This is not the same as abandoned — HEY reports what it can observe, nothing more.

Resumed building

Shipping again in the last 14 days after a gap of 60 days or more. The gap is measured from the comeback itself — everything shipped inside those 14 days is the comeback — back to the last update before it, so a project that ships twice on its first day back still reads Resumed. Only a gap HEY watched counts (since hbm-v25): every ship of the comeback must come from a source HEY was already reading when the gap began. A repository HEY starts reading during the gap, or a ship whose source HEY cannot name, is not a return — the project reads Shipping or Active instead.

Activity unknown

HEY has a source or two but not enough it can currently read to say either way.

No builder source linked

HEY has no repository, changelog or feed linked for this project, so whether it is building is not measured. It does not mean nothing is being built. Trading is not building. HEY looked and found nothing to read building from, which is a different statement from Dormant: that one means HEY looked at a source and saw no work.

Activity not measurable

HEY has a ship on record but no repository, changelog or feed it can keep reading, so whether this project is building now cannot be judged. The last ship is the newest evidence HEY holds. A Verified Builder badge beside it is not a contradiction: the badge is about what was shipped, the status about whether HEY can see it shipping now.

How the counts nest

Activity is never derived from token price, market cap, trading volume, holders or wallets.

HEY Build Momentum

A 0–100 measure of how consistently a project ships meaningful work, combining:

• 35% shipping recency
• 25% shipping consistency
• 20% shipping significance
• 15% verification quality
• 5% source diversity

Bigger deliverables carry more weight — a product launch counts 10 where a documentation update counts 3, and an announcement counts 0. Repeated updates of the same kind have diminishing returns, and aggregated commit activity is capped, so neither commit volume nor marketing volume can inflate the score.

Discovery Gap and Still Building

Discovery Gap compares how actively a project ships against its current market footprint. Both percentiles are taken over the published projects that have a market reading — a live one, that is the project's own. A token with no live market has no Discovery Gap and is not in the population the others are compared with. Nor does a token that nothing the project publishes ties to it: when the token is not Verified, HEY knows it only from a market listing rather than a launch on this chain, and every website and repository HEY holds for the project is only context — a bridged copy of another chain's asset, for instance — the local pool says nothing about the project's market, so HEY does not measure it. A market that is live but not active — low liquidity, no trades in a day, readings HEY cannot yet settle, or only a launch curve — stays in that population, but gets no Discovery Gap of its own and cannot be Under the Radar: its attention sits near the floor, so its gap would measure the thin market rather than the building. HEY says “not measured — market too thin” there, never a number. The same holds for a token whose only market is its launch pool (since hbm-v22): its “liquidity” is its own supply sitting in the pool that launched it, at its last price, so it is labelled “Launch pool only” and gets no Discovery Gap, no Under the Radar and no Still Building. A market that is active only because another pool of the same token holds the liquidity is measured on that pool's reading — its valuation, liquidity and volume — never on the thin pool the token's own reading describes; when HEY holds no current reading of that pool, it says “not measured” instead. A project appears Under the Radar when the scoring rule finds it eligible — it is active, has Build Momentum of at least 30, has at least 2 source-backed updates in the last 30 days, at least one of them a ship rather than a commit summary, has fresh market data and a positive Discovery Gap, because a gap of zero is not a gap. The lists apply the same positive-gap test again when they read the rule's answer.

Under the Radar also needs the project's market attention to be typical or lower (since hbm-v30): its market-attention percentile below 60, never elevated or high. A positive gap beside high attention is still published as a gap, but building ahead of attention is not claimed for a project the market already notices. The population the percentiles are taken over does not change.

Nor is a Discovery Gap measured on a token HEY cannot tell from a lookalike (since hbm-v30): when the token is not Verified and another published project's token, on a different contract, carries the same ticker, HEY says “not measured — the token is not tied to the project”, and the project cannot be Under the Radar. Its market stays in the population the others are compared with.

And a Discovery Gap needs a market reading from the last day (since hbm-v30), the same freshness Under the Radar and Still Building ask for. With only an older reading HEY says “not measured — no current market reading”; that reading still counts in the population the others are compared with.

Nor does a project whose activity status is Unknown — building HEY cannot read — get a Discovery Gap or appear Under the Radar (since hbm-v27): set against the market, momentum with no readable recent ship would claim building ahead of attention where HEY cannot read the building. HEY says “not measured — activity unknown” there, never a number, and the project stays in the population the others are compared with.

A valuation HEY cannot stand behind is no input. It is withheld — shown as withheld with its reason, never as a figure — when it is at least 10,000× the liquidity measured in the same reading; above $10B on a Robinhood Chain token that no listing HEY reads carries; more than 10× away from HEY's own chain readings of the token's price; or more than 10× away from another source's price for it within a day. A withheld valuation leaves the Discovery Gap's market percentile — the project's liquidity and volume are weighed as for any missing figure, and the value leaves the population the others are ranked against — and it is no current value and no tracked high for Still Building, which then reads “not measured”, never “not met”. A figure HEY could not read never trips the gate, and the gate only withholds: it never awards a badge and never reaches activity status or Build Momentum.

Still Building is shown when a project's market value is at or below 50% of the highest daily close in HEY's price history for the token over the last 90 days — HEY's own readings and the daily price history it stores from DEX data providers, which reaches back before HEY began reading the token — that high is at least 7 days old, the project has at least 2 meaningful updates in the last 30 days, at least one of them a ship rather than a commit summary, and meaningful building continued after that decline began. A daily close counts from the end of its day, so an update published earlier the same day is not building “since” the decline. Under the Radar is never given where the Discovery Gap is not measured, and Still Building never where the token has no live market of the project's own. Its drawdown is measured on the same markets as the Discovery Gap: on a market that is not active or is only a launch curve, a “decline” measures the thin market, not a market that fell while the team kept building, so HEY says “not measured — market too thin” there rather than giving or refusing the badge. The same holds for a project with no tracked token, no current market reading, or building HEY cannot read: Still Building is not measured there, which is different from not held.

These are factual observations about past activity and market context. They are not predictions, not valuations, and not investment advice. HEY never claims that continued building will affect a token's price.

Token market status and token verification

A project can be building while its token has no market, and a token can trade while nobody builds. HEY keeps the two apart. Activity status reads ships and repositories. Token market status reads HEY's own market snapshots and is shown beside it, never inside it: Active market, Low liquidity (under $5K; a market already low stays low until it reaches $6K, so a pool hovering at the line does not switch back and forth), No liquidity, Trading inactive (liquidity but no trades in 24 hours), Liquidity no longer detected (HEY recorded at least $5K earlier and now sees a tenth of it or less, under $25K), Market not detected (no reading for 30 days after a market existed), or Market data insufficient. These are observations HEY made; none is a verdict on why.

An active market needs a measured volume. When the newest reading reports no 24-hour volume and no fresh trade record does either, HEY uses the newest volume any reading measured in the last 36 hours; with none, a pool that holds liquidity is Market data insufficient (“Readings not confirmed”), never an active market and never a volume of zero — whether it traded is not known.

A 24-hour volume a source repeats unchanged, beside an unchanged price, for more than 36 hours is a frozen feed, not a measured volume, and HEY reads it as no volume. A depth reading older than 7 days makes no market active, whatever trades say since (“Readings not confirmed”). A Pons launch that graduated into a two-sided pool, seeded with the token and a paired asset, is judged as a market once HEY reads that pool; only a pool still on its curve, or seeded with no paired asset, is a launch pool. The highest liquidity HEY reports for a token is the level it held on two consecutive daily closes, never one reading.

Market data insufficient also covers readings HEY will not believe. A reading from another pool counts only if its price is within 3× of HEY's own price for the token. A large pool that almost nothing traded in, and that HEY's own reading of the chain does not find, is shown without its liquidity or valuation. Neither case, nor two readings that disagree, counts as a live market for a badge.

Token verification asks a different question: is the token HEY tracks the project's own? Finding a token through a launchpad record, a market listing or a matching name establishes nothing. A token is Verified only when the project itself ties the contract to the project: its own site names the address, or the owner proves control of the deployer with an on-chain signature. Contract mismatch means the project's site names a different contract than the one tracked. Everything else stays Unverified, with the reason on the page.

The links on a project page carry their own, separate label, and it is not the same question as either of the above. Verified means HEY confirmed the project controls that link. Official means HEY found it published on the project's own domain, which is evidence rather than a claim. Declared by the project means someone put their name to it on a submission and a moderator approved the page — a claim with a person behind it. From a launch or market listing means HEY copied it from a launchpad form, a DEX profile or a token's market listing: what the token's record said, not the project speaking, and when the project's own site links a different account, the site wins. Robinhood's own accounts and sites are never a project's, whatever recorded them. Unverified means HEY found the link and nobody has stood behind it. A project can be a verified builder while its links are only declared: the badge is about what was shipped, the label is about who says the link is theirs.

Two consequences. Still Building needs a market to decline against, so a token with no liquidity, a removed market or a launch pool nobody traded never carries the badge. And the home page lists projects whose tracked token still has a market, or no token at all; the rest keep their page and their place on Explore, with the fact stated. Neither changes activity status or Build Momentum.

Evidence

Every update keeps its source. HEY distinguishes what a project claims from what can be checked: self reported means a builder told us, source linked means it came from a source registered to that project, and source verified means HEY confirmed it against a public source such as a GitHub release. Disputed and retracted updates keep their record but stop counting.

A package published to a registry such as npm, PyPI or NuGet is never a ship on its own. The release in the project’s repository is the ship; a registry copy of it is kept as context and counts toward nothing.

Scan evidence breakdown

Model scan-evidence-v1. A scan reports how much of a contract HEY could actually verify from public sources. It measures HEY’s evidence and nothing else: it is not a measure of a project’s quality, it says nothing about what a token might do, and a low reading means HEY could see little, not that there is little to see.

Who built it

How firmly the contract is tied to a named builder. The two checks that actually establish a tie — the project’s own site naming the contract, and the repository being established as that contract’s — carry most of the weight. A declared link counts for little; a corroborated one counts for a lot. A name that merely matches counts for nothing.

What is being built

The project’s stored Build Momentum, unchanged. A scan never computes a development number of its own, so this is the same figure the project page shows. A contract HEY holds no recorded ship for has no Build Momentum, and the band says so rather than reporting zero.

What HEY can see

Of the places HEY looks — the chain, the explorer, the creation record, the decoded calls, a declared site, a repository, a listing profile and HEY’s own ship records — how many answered. This one is about HEY rather than the project, and the blind spots beneath it name every place that did not.

Overall evidence

45% who built it, 35% what is being built and 20% what HEY can see, renormalised over the bands that could be measured. Identity outweighs build deliberately: build evidence read through a repository nothing ties to the contract is evidence about somebody else’s work, so a weak identity tie also holds the overall wording back from “strong”. Fewer than2 measured bands reports no overall figure at all.

The scan prints the word only, never the number behind it (2026-10-09). Readings are 40–69 partial evidence and70 and above strong evidence; below 40 is limited evidence. Each is shown with its word as well as its tint, and none of the three is red — red is reserved for real errors, and thin evidence is not an error.

A check HEY could not run leaves the total as well as the tally, so a provider that did not answer costs the reader a named blind spot and costs the project nothing. A check HEY ran that came back empty does count, because HEY looked. That is the scoring form of the rule that absent means unknown, never zero.

For a token HEY does not track, a scan also reads its on-chain context (2026-10-09): how many addresses hold it, the Top 10 and Top 50 shares of its supply with pools, lockers and burns left out, and the transfers, trades and distinct addresses of the last complete UTC day and of today so far. Shares and counts only — never an address, never a person. It sits after the builder groups, is not part of the evidence breakdown above, and a part HEY did not read says “not read” with its reason, never zero.

Peer context

A figure on its own says little, so HEY places some of a project’s figures among comparable projects — rules version peers-v1, recomputed once a day. A comparison is made only when all of these hold:

• Same type. The group is the project’s primary narrative at its top level (DEX, Lending, AI, …) and its kind: memes are compared with memes, everything else with products. A memecoin is never compared with a lending protocol. A project with no narrative, or a narrative that says where it is rather than what it is, has no group.
• Same definition and window. One written definition per figure; Build Momentum only under the current scoring version.
• Measured only. A project whose figure HEY did not measure is left out of the group, never counted as zero.
• Enough projects. A median and a p10–p90 range need 8 measured projects; a percentile needs 20. Below that HEY says “not enough comparable projects”.

Build Momentum

The published Build Momentum under the current scoring version, for researched projects whose building HEY measured. Shipping evidence only; no market figure reaches it.

Meaningful events · 30 d

Meaningful builder events in the last thirty days, counted as the project’s velocity counts them (a week of code activity is one event).

Days between releases

The median number of days between release days over the last year; measured once a project has at least three release days.

Active contracts · 7 d

Watched contracts that took at least one call over the last seven complete days, for projects whose usage is measured on every one of those days.

Contract calls · 7 d

Calls into the project’s watched contracts over the last seven complete days, for projects whose usage is measured on every one of those days.

Protocol TVL

DefiLlama’s TVL on Robinhood Chain for the one protocol the project is matched to, on the registry’s newest day. A project matched to several protocols is not compared.

Protocol fees · 24 h

DefiLlama’s fees over 24 hours for the one protocol the project is matched to, on the newest economics day, where DefiLlama tracks them.

Protocol revenue · 24 h

DefiLlama’s revenue over 24 hours for the one protocol the project is matched to, on the newest economics day, where DefiLlama tracks it.

Each figure stands alone. HEY never combines them into one score, “above the median” is a position rather than praise, and no peer comparison feeds activity status, Build Momentum, the Discovery Gap or the Radar. Research completeness is not compared: coverage is a set of states, never a score.

Relationships

For each project HEY lists what it is connected to and why: its token, its contracts and where a proxy points, its official repositories, domain and docs, the packages tied to it, where it launched, the registry that lists it, the independent readings that confirm its identity, relationships between projects that a moderator verified, and the audits and bug-bounty programs its security context holds — each labelled by where it is published, as evidence that an audit took place and never a verdict. Every relationship carries its standing — verified, official, claimed, context only or observed — the records behind it and when HEY observed it.

No account is ever part of it: no deployer, holder, caller or wallet. HEY draws no partnership, because nothing it reads proves one, and an on-chain interaction between two projects’ contracts is a different claim that HEY does not hold yet — it says so rather than implying either.

Reading HEY as data

Everything on these pages is also readable as JSON, with no key and no account: /api/projects for the catalogue, /api/projects/{slug} for one project, and /api/ships for what shipped. The same filters as the browse pages, the same counts, and every fact carries the source behind it. Anything HEY does not know is left out rather than sent as a zero. For a reader rather than a script, the same material is published as RSS.

What HEY does not do

No wallet analytics, no PnL, no smart-money or whale labels, no wallet profiles or clustering, and no “follow this wallet” surface. HEY does not build a picture of an address across tokens or over time.

One narrow exception, since September 2026: the market page of a single token shows a distribution map of that token's largest balances — how concentrated the supply is, and which balances are a pool, a locker or burnt. An address there is a point on a chart of one token's supply. It is never a person, never scored, never ranked across tokens, never followed, and never an input to activity status, Build Momentum, the Discovery Gap or the Builder Radar. The source HEY reads it from reports only balances that moved in about the last nine days, so for a quiet token the page says “No balance change in the provider's window” and keeps the last map HEY read, with its date: an empty answer never means nobody holds the token.

HEY does not rank by price, and organic placement cannot be bought.