SAFU. RISKY. RUG. Blue Chip. What each verdict actually means.

The verdict is a vibe, not a verdict

Every scan ends with one of four words on a card. Sometimes a badge on top. That card is the entire point of this product, so it's worth knowing what we mean when we put a label on it.

The short version: we read public on-chain data, score the token from 0 to 100, the score lands in a bucket, and the bucket gets a name. That name does not promise anything. It's a summary of what the data looked like at the moment you pasted the mint. Ten minutes later it could read different. An hour later, very different.

I'm a doomer wojak in a basement, not your fund manager.

With that out of the way, here's the actual breakdown. One tier per section, in order of how mad your wallet should be.

The cope score, briefly

Before we get to the labels: the number on the right side of every verdict card is the cope score. Zero is chill. One hundred is pretty much already rugged. The bar fills accordingly.

It's a weighted sum of every signal we track, and the score lands in a bucket:

• Low score, no high-severity signals lands in SAFU.
• Medium score, or one or two stacked mediums lands in RISKY.
• Anything with a high-severity hit, or enough mediums to look intentional lands in RUG.

The weights aren't pulled out of a hat. They get re-fit against real outcomes every week. If a signal stops predicting, its weight shrinks. If a new pattern starts blowing up wallets, it gets added in shadow first, scored against actual rugs, and only starts counting once it earns the right to.

So: word is the bucket. Number is the score. Let's name the buckets.

SAFU (low risk)

Checked it. Nothing screaming. Clean as these get.

SAFU is the green tier. Low cope score. No high-severity signals tripped. Liquidity isn't a single wallet pretending to be a pool. The holders aren't ten alts of the same person. The dev hasn't done anything that screams. Mint authority and freeze authority are revoked or held by something that doesn't look hostile. The structural box is checked.

A SAFU card mostly comes in two flavors. The first is fresh-and-clean: a brand-new token that did everything right at deploy. Authorities revoked, LP burned or locked, no sniper bundle on the first slot, no dev wallet stack of zeros behind it. Nothing structural to flag. The chart could still go any direction. SAFU just says no one rigged the box before they handed it to you.

The second is mature-and-clean: a token that's been trading for weeks, holders have spread out, liquidity is real, no one's done anything dumb yet. This one usually picks up the Blue Chip badge too (more on that below).

Things SAFU does not mean:

• The chart will go up. SAFU is silent on direction. It speaks only to structure. A token can be perfectly clean and still bleed to zero because nobody buys it. Clean and dead are not opposites.
• The team is doxxed. Most Solana tokens have anonymous teams. SAFU does not require a team page or a LinkedIn. If you need a team, you have to check that yourself.
• It can't rug later. SAFU is a snapshot. Devs wake up. Wallets get unlocked. A token can read SAFU at 5 minutes and rug at 50. Rescanning before you size up isn't paranoid; it's free.
• The risks list is empty. Scroll down on a SAFU card and you'll usually still see low and medium flags. SAFU just means none of them rose to the level where we'd call it on you.

Read the risks list. Read the score. SAFU isn't a finish line; it's a vibe check that came back positive.

RISKY (medium risk)

Yellow flags ngl. Seen this one before.

RISKY is the orange tier. Something flagged. Usually more than one thing. Nothing severe enough on its own to call it a rug, but the pile is real and the pile is what gets you.

The combinations that land here keep showing up:

• Liquidity is thin or unlocked, but holders look fine.
• Holders are concentrated, but the contract itself is clean.
• The token is brand new and we just don't have enough age, volume, or holder count yet to commit either direction.
• Bundlers and snipers ate a noticeable share of the supply at launch.
• A small but non-zero overlap with wallets that have rugged before.
• The LP is locked, but only for 24 hours, which is not really locked; that's just a delay.

RISKY is the most honest tier we have. It's where the scanner says "I don't love it, but I'm not gonna lie to you and call it a rug." And it's the tier where calibration matters most: on metrics you can see the share of risky-bucket scans that resolve as rugs at +24h. It isn't zero. It isn't one hundred. It's the bin where reading the actual risks list before you make a call either way is, structurally, the most valuable thirty seconds you can spend.

A useful mental model: SAFU means there's nothing to read carefully. RISKY means there absolutely is.

RUG (high risk)

Oh no. I can already see the chart.

RUG means at least one high-severity signal is on, or several mediums have stacked into something that looks like a launch engineered to fail. The genre catalog, by frequency:

• Mint-still-on rug. Mint authority active. The dev can print supply at will. The entire premise of "fixed supply" is a lie.
• Freeze-still-on rug. Freeze authority active. The dev can freeze your wallet. You become a passenger.
• LP-pull rug. Liquidity not locked or burned. Yanking it is one transaction. The chart goes vertical the wrong way.
• Bundle-and-dump. Snipers and bundlers ate a huge share of the float in the first block. The people holding most of your token came to flip, not hold. The candle to zero is patient but inevitable.
• Familiar-funder rug. Dev wallet funded by a hot wallet that has launched eight other tokens this week, all currently zeros. The pattern repeats because it works.
• Coordinated-holders rug. The top wallets show up funded by the same source on the same day with the same buy size. One team in five trench coats.

We call it RUG even when, technically, the token has not rugged yet. That's the point. By the time the chart actually nukes, it isn't a verdict; it's an obituary. The label is meant to be useful before the event, not after.

Does this catch every rug? No. Some are hard:

• The slow rug, where the dev waits a week or three before pulling, partly to let calibration scoring forget about them. We still catch most of these via the wallet history layer, but not all.
• The social rug, where the contract is pristine and the team disappears off Twitter. On-chain, there's nothing to see. That's not our layer.
• The migration rug, where the token graduates from a launchpad to a real pool and the original liquidity gets stranded. Mostly caught now. Occasionally not.

We publish the miss rate on metrics. It is a real number. It is not zero. We are not pretending to be magic; we're trying to give you a few percent of edge on average. That compounds.

Blue Chip (the badge that sits on top)

Ok ser. This one survived cycles you didn't.

Blue Chip is not a fourth tier. It is a badge that gets pinned onto SAFU when the token also clears a maturity gate. Roughly:

• Enough age that there's actual on-chain history to read.
• Enough liquidity that one whale can't move it 30% by sneezing.
• A real, distributed holder count.
• Nothing high-severity in its lifetime, not just at scan time.

The verdict label still says SAFU underneath, because we want the metric matrix to stay three-tiered and the calibration cohorts to stay clean. The badge just tells you "this isn't a fresh shiny thing; this one has paid its dues." Boomer DD.

What Blue Chip is not:

• A guarantee the chart goes up. A Blue Chip can still bleed 80% in a cycle. That's a price problem, not a rug problem.
• Permanent. A token can lose the badge if its liquidity dries up or the holder distribution collapses. The criteria are checked on every scan, not stamped once.
• A measure of cultural relevance. Structure ages well; memes don't. Blue Chip says structure is intact, not that anyone's still trading it.

If you're looking at a Blue Chip card and the cope score is creeping up, take that more seriously than you'd take the same score on a fresh launch. Mature tokens that start to flag tend to do so for real reasons.

What none of these tell you

How we know we're not just making it up

Every scanned token gets followed up. We snapshot it at +1h, +6h, +24h, +72h, and +7d after the scan and line up what we said with what actually happened. That's the calibration loop, and the per-tier accuracy lives on the metrics page.

We publish the misses on purpose. We do not retroactively edit old verdicts to make ourselves look better. Once the label is pinned, it's pinned. If SAFU starts rugging more than it should, the weights re-fit. If RUG starts saving more than it should, same. The rule version on metrics tracks every iteration; you can watch each tier's behavior change over time.

If any of this ever stops being true, you'll see it there before anywhere else. That's the deal.

TL;DR

Four words, one number, one badge. The word is the bucket. The number is the score. The badge is the pedigree. All three are summaries of public data, calibrated against what happens to similar tokens. None of them are a substitute for actually thinking about what you're buying. You already knew that.