The operator's field manual for SniperIntel.

SniperIntel is a private intelligence product for Pump.fun developer wallet tracking and scoring. It turns the continuous noise of new creators into a scored, filtered, context-aware research surface for Solana operators.

In practical terms, the product watches the wallets that launch tokens on Pump.fun, classifies them, assigns a ranking score, tracks their history, measures how crowded their setups are, and validates all of that against measurable outcomes. It is not a wallet explorer, not a generic dashboard, and not a trade-execution platform.

You are meant to use it as a pre-trade research filter: before you size anything, you ask the product who the developer is, how they have behaved historically, whether anyone else has already crowded the setup, and whether the broader population of wallets in this score band actually performs well over time.

SniperIntel is a public ecosystem of five connected surfaces. Each one has a job, and the jobs don't overlap. Knowing where you are in the system saves you from reading the wrong page for the wrong question.

02.1Reading order for a new customer

• Home — positioning and product entry point. Skim, then move on.
• Methodology — the scoring framework itself, optional before the guide.
• Guide — this page. Deep operating manual for how the dashboard actually works.
• Results — proof of outcomes. Read after you understand the scoring logic.
• App — the live dashboard. The rest is context for what you see here.

The dashboard is built around a repeatable operator loop, not around dashboards for their own sake. The same loop runs whether you are evaluating a single developer, sweeping a classification, or cross-checking a setup you already saw on-chain.

The loop is deliberately boring. Boring is what lets the product stay honest about what the data supports and what it doesn't. The interesting work is still yours — the product just makes sure you are looking at the right developer, in the right context, with the right history.

Each tracked developer wallet gets a composite score that represents ranking and context, not a buy signal. It summarizes historical behavior and the competitive environment around the developer, and it places them against the rest of the tracked database. The six publicly explained dimensions that feed it:

04.1What stays proprietary

The six dimensions above are public. The weights, threshold tables, and the order in which rules stack are not. This is intentional: a fully-exposed rule set becomes a commodity and gets gamed at the creator-wallet level within days. The public Methodology page explains which signals go in and which limits apply, nothing below that.

Every wallet in the database is tagged either ST or MT. This is one of the most important interpretation rules in the product, and also one of the easiest to miss. It directly changes how much confidence you can place on a score.

Practical rule of thumb: if you see a high-scored ST wallet, read it as "this single token looked clean" — not as "this developer is reliable." An MT wallet with the same score is making a stronger claim because the pattern has repeated. New developers all start as ST. Some will never cross into MT because they never ship a second token.

The wallet feed is the primary discovery surface. It's where you land when you open the app and where you start any session. It lists tracked developer wallets with their score, classification, trend, and status — and it is the thing that turns "the whole of Pump.fun" into "the small list you actually want to look at."

06.1How to use the feed

• Filter before you scroll. Start with classification, a score floor, and trend. Scrolling the entire database is noise.
• Search by address when you are cross-checking a wallet you saw somewhere else. The feed is the source of truth for what we track.
• Watch the status tags. Active, watch, stale, and avoid are meaningful signals — not decoration.
• Don't ignore locked preview rows. They exist so you can see what's out of your plan's visibility scope without pretending those wallets don't exist.

Wallet detail is the drill-down view for a single developer. It is where the composite score stops being a number and becomes a readable behavioral picture. Everything you need to decide whether to keep watching the wallet or drop it lives on this screen.

07.1How to read this screen

• Start with the dimension breakdown, not the headline number. A composite 92 with a Competition subscore below 0.65 is telling you the setup is crowded — the top-line score alone hides that.
• Token history is the sanity check. Repeated graduations with clean fee structure is a pattern. A single big ATH surrounded by stalled launches is noise.
• Score history shows direction. A rising slope over 30–90 days is a developer stabilizing. A declining slope is a developer you should consider downgrading in your own list.
• Locked previews for out-of-plan wallets will show a partial profile with the detailed screen gated. That is by design — it prevents cross-plan data leakage.

Competition is not simply "how many bots showed up." A developer with strong scores can still be a bad trade because the setup is already saturated at entry. The competition surface exists so you can tell the difference between a clean board and one where the edge has already been priced in.

08.1What the competition surface actually shows

• Sniper density — the raw count of active snipers on the developer's tokens.
• Buyer-set composition — which known bots are in the buyer set, and how consistently.
• Fee pressure — how fees compare against the rest of the database; crowding pushes fees up.
• Bundle detection — structural buy bundles that can poison an otherwise clean setup.
• Per-token breakdown — the same metrics split across the developer's last few tokens so you can tell crowding direction.

Proof is where the product validates itself. It exists so that every claim the scoring framework makes can be checked against measurable outcomes — not marketing copy, not anecdotes, not screenshots. If the product is right, the numbers on this surface should hold up over time. If they don't, the framework owes you an answer.

• Launch rate & graduation rate — how much the platform actually produces, and how much survives.
• Score accuracy — the correlation between the score band a wallet was in and the outcome its tokens actually produced.
• Cohort analysis — score bucket performance, so you can tell if A-band wallets meaningfully outperform C-band wallets over time.
• Live tracking — the current monitored wallet count, which anchors all of the above in a population, not a sample.

You should look at proof before you trust any single wallet score. A wallet in a band where the cohort-level outcome rate is mediocre gives you different information than a wallet in a band where the cohort-level rate is high — even if both letters look the same.

Linking your wallet turns on read-only personal analytics. That is the entire scope of the feature. The product can see what you did and compare it to the platform; it cannot move funds, sign transactions, or execute trades.

10.1Why you might want to link

• Execution sanity check — see your own fees and slippage vs. the platform average.
• Cross-reference — spot which tracked wallets you already interacted with and at what score band.
• Personal cohort — review how your own trades distribute across A/B/C score bands over time.

Bot intelligence is the surface where you inspect the mechanical buyers in the environment — the wallets that consistently show up early, across many developers, with recognizable behavior. It is not a morality page; it is a context layer for understanding who else is in the room.

• Bot directory — classified mechanical buyers with visible behavioral fingerprints.
• Timing profiles — typical entry windows, hold durations, and exit patterns per bot.
• Fee behavior — how aggressively a given bot pays up, which informs whether you can compete on fees at all.
• Overlap with competition — which bots are showing up on the developer you are currently inspecting.

In practice, you read bot intelligence together with the competition panel on a specific token: if a setup is crowded and two of the top bots are ones that consistently pay 50%+ fee premiums, the math for a manual entry is already bad before you click anything.

Analytics is the broader pattern-exploration surface. It is where you look for platform-level relationships rather than single-wallet detail. Not every user needs it on day one — but every serious operator ends up here eventually, because this is where structural questions get answered.

• Score ↔ outcome relationships across the full tracked population, not just cherry-picked wallets.
• Classification distribution — how the ST / MT / verified / risk mix changes over time.
• Behavior shifts — when fee structures, crowd density, or graduation rates move across the database.
• Cohort drift — how last month's A-band is performing now vs. this month's A-band.

Use analytics when something feels off at the single-wallet level and you want to know if the oddness is a wallet problem or a platform-wide problem. The answers here don't affect a single trade directly, but they affect how you interpret every trade for the following weeks.

Some surfaces are plan-scoped. This section exists so the product logic is legible, not so the page becomes a pricing table. If you want prices, use the pricing section on the homepage. Here we only describe what changes in the product depending on your plan.

13.1How to read locked previews

When a wallet is out of your plan scope, you still see it in the feed — as a locked preview. The row shows the classification and a blurred score band, with the detail screen gated behind upgrade. This is intentional: pretending out-of-plan wallets don't exist would distort the research surface. Preview rows are visible; deep inspection is not.

Short answers to the questions that show up in support most often, and to the misconceptions that cost new customers the first week.

A short and deliberately blunt list. Carry these into every session with the product and you will save yourself a month of bad assumptions.

• It does not execute trades. Ever. Not today, not when linked wallet is on.
• It does not custody funds and it does not ask for signing permission at any point.
• It does not predict outcomes. It ranks wallets against a database and tracks how those rankings perform.
• It does not guarantee edge. Crowding, fees, and market conditions can erase edge even on top-scored wallets.
• It does not expose weights or thresholds. The six dimensions are public; the stack is not.
• It is not a generic explorer. Address search is for cross-checking tracked wallets, not for browsing every wallet on chain.
• It does not replace your own research. It narrows the surface so your research is pointed at the right wallets.

You have the manual. Pick the next step based on what you came here to do.