API
Show your users where a token came from
Your user is looking at a token they have never seen, about to act, and asking what it is. The honest answer available to every wallet and scanner today is a price chart, because everything readable on chain was created minutes ago: the wallet has no history, the mint is new, the allocation is ordinary. There is nothing in it to recognise.
{
"birth": { "capturedAfterSec": 1.36, "devPct": 1.74 },
"signals": [
{ "code": "image.crossCreator",
"creators": 3,
"evidence": "/api/v1/image/…" },
{ "code": "creator.repeat",
"priorLaunches": 14,
"graduatedConfirmed": 2,
"evidence": "/api/v1/creator/…" }
],
"outcome": { "graduatedConfirmed": true }
}One call changes the answer. Pass a mint and the response carries that launch's own record and what it shares with every other launch we hold: the identical document, the identical picture, the ticker, the creator's other launches, each as a count with the hash behind it. Your product can say that this token declared the same file as hundreds of earlier launches, in the second it appears. It is not a rare case: one in five of every creator recorded launched with a file another wallet also launched with, and 391 ticker symbols were each carried by fifteen or more separate mints in 22 days.
What "the identical file" means in a response
{
"name": "BRAIN",
"symbol": "BRAIN",
"description": "Scroll less. Think faster. Ape smarter 🧠 $BRAIN turns timeline chaos into a notice-think-ape loop built around conviction instead of noise.\nCreator rewards feed planned buybacks, burns, and on-chain receipts while organic frens rally around @brainclub_fun without bot-farmed hype 📲",
"image": "https://ipfs.io/ipfs/bafkreiawbz2wzxobciazym4rde6uj6374odpntcyfliz6yoh6af55hqdeq",
"showName": true,
"twitter": "https://x.com/brainclub_fun/status/2104290425922040081",
"telegram": "",
"website": "https://www.apeonbrain.fun",
"banner": "https://ipfs.io/ipfs/bafkreie4afuv7bgy3elmb2it5zecuofcso5pr2at53aw4yk3e7sa34fsui"
}| Wallet | Token | Launched | Fingerprint of its document | |
|---|---|---|---|---|
| 1 | 6459…jPd1 | 2svC…GJWF | +0 ms | 6151611e…e77d04 |
| 2 | 9L9x…bDzX | CscP…8t7m | +6 ms | 6151611e…e77d04 |
| 3 | 4zDw…22QB | 79LC…aoNr | +17 ms | 6151611e…e77d04 |
| 4 | BDc1…efRG | 43XG…oyj1 | +18 ms | 6151611e…e77d04 |
| 5 | 4rgY…vYy1 | 4BvL…V8RP | +19 ms | 6151611e…e77d04 |
| 6 | Crmq…zG6n | 9TY8…AmbV | +19 ms | 6151611e…e77d04 |
| 7 | 8Hzg…1tkt | GWKs…RQDf | +22 ms | 6151611e…e77d04 |
| 8 | hwGG…ky9S | 5ZXD…f9DX | +23 ms | 6151611e…e77d04 |
| 9 | AFNh…vJdi | Az55…gS5Q | +27 ms | 6151611e…e77d04 |
| 10 | A2L1…wVyC | 33WT…PWoj | +28 ms | 6151611e…e77d04 |
| 11 | Do2o…hpjk | EKnp…b7XL | +28 ms | 6151611e…e77d04 |
| 12 | 5zEr…pM4B | 285H…EZoC | +29 ms | 6151611e…e77d04 |
| 13 | DjNv…hS2c | E9JW…asb1 | +31 ms | 6151611e…e77d04 |
| 14 | DcLr…5Q3H | 4rus…68cC | +34 ms | 6151611e…e77d04 |
| 15 | DgzE…ugN6 | EFZa…AaHc | +37 ms | 6151611e…e77d04 |
9d34004a28966075d4b02572e615c3da5339283d89803d849d37de4afd8ad45binstead of 6151611e…e77d04. Nothing in common.A real launch document from the record, 27 September 2026 (UTC), printed as it was captured. Wallets and tokens are truncated. Times are our capture clock; a copy of every one of the fifteen was held within six seconds of its launch.
The match
Every launch publishes a small file about itself: its name, its ticker, a description, its picture, its links. Above, fifteen tokens from fifteen wallets carry one: the same 281-character description, the same emoji, the same links in the same order, 694 bytes and not one of them different. A fingerprint is computed from every byte. There are about 1077 of them, and no two different files have ever been shown to share one. These are not fifteen similar files. It is one file, fifteen times.
The line it draws
A wallet costs nothing and takes a second to make, so a new one arrives with no history, and money can be routed so that no transfer ever joins two of them. The file is the one thing that gets reused, because launching again quickly means reusing it. So every match draws a line between two wallets that nothing on the chain connects.
What the lines add up to
Lines join. One wallet shares a file with a second, the second shares a different file with a third, and three unrelated wallets become one group, with every launch any of them made. A thousand wallets stop being a thousand strangers and become a handful of groups that can be counted, dated and followed.
Why this is the whole thing. Without the file, every launch is a stranger's first. With it, counsel is looking at one group of wallets and not a thousand unrelated ones; a brand is looking at one run of launches and not hundreds of separate incidents; and a wallet app can tell its user that this exact file has launched before, in the second the token appears. None of it can be rebuilt afterwards: the files stop being served, and a fingerprint can only be taken of a file somebody kept.
Made, not stumbled on. A name is reused over weeks. A file is reused within minutes: of 85,952 groups of launches carrying one identical document, 81.83% were complete inside five minutes. The one route to identical files with nobody copying, a launch tool's blank default, is the kind shared least. What a match means, measured 2 September to 19 October 2026.
How a match leads to who. A match establishes that the same file was used, and a connection is never drawn from one coincidence. The person behind a wallet is known to the exchange or service its money came through, and a subpoena is how that is obtained. The provenance report traces each wallet's funding upstream hop by hop, with the transaction behind every hop, so counsel knows who to subpoena.
You could not build this, and that is not a sales line. The files a launch publishes about itself sit on hosts their creators control and can stop being served at any time, so there is nothing to backfill from: an indexer started today sees none of it. We have been capturing at the creation transaction since 2 September 2026. What you get here is the record and nothing else - we never tell your product whether a token is safe, because then you would be defending our judgement instead of showing a fact.
Every record on this site is also a JSON document, served to API keys. Send the key as Authorization: Bearer <key>. Plans and enterprise agreements. /api/v1/status stays open, and so does every token page on this site.
There is one thing you have to get right, and it is the next section.
Absence of a record is not a clean record
The field to branch on is finding.class, which is one of FINDINGS, NOTE, NO_MARKERS or UNKNOWN. These name classes of entry on a record, not grades of risk and not advice. We record what a launch was; what that is worth is yours to decide. Read them as:
| FINDINGS | a finding: something we recorded about the launch, such as the creator holding most of the supply, or no outside wallet buying its curve |
| NOTE | a note: recorded, weaker, and often about the present rather than the launch |
| UNKNOWN | not established. Not a mild warning: an absence. We could not settle the question |
| NO_MARKERS | we checked this launch against every marker we record and found none. A statement about what we checked, not an endorsement of the token |
cleanAtBirth is a convenience and it has three states, not two:
| true | we watched this launch (or rebuilt its complete history) and checked it against every marker we record; none is present |
| false | we watched it and at least one marker is present, or one criterion is unmet: see finding and flags for which. Not an accusation: a token whose pool we could not read just now is false and UNKNOWN, not FINDINGS. |
| null | we do not know. We did not observe the launch and have not rebuilt it. Do not render this as clean, safe, or "no issues found." |
Once a token's float has been spread across wallets, a launch that was assembled and one that was not look the same to present-tense inspection, and that is the entire reason this archive exists. A null means the evidence is gone, which is the opposite of reassuring. Every refusal and every error we return also carries finding.class = "UNKNOWN", so code that reads only that field is safe even when it ignores the HTTP status.
Endpoints
| GET /api/v1/token/{mint} | one launch record: what the creator took in the first block, how many outside wallets bought its curve, how it graduated, who took it, and the pool right now |
| GET /api/v1/wallet/{address} | a wallet's priors: every bonding curve it has bought outright in this archive, and what it did with the tokens afterwards. A wallet we have never seen returns inArchive: false and nulls, not zeros, because "we hold nothing on it" is not "it has done nothing". In buyouts, sameBatchAsLaunch: true means the buy arrived in the same batch of chain events as the launch itself, hoursAfterLaunch is then null rather than 0, because our timestamps cannot resolve it further. Don't render it as zero. |
| GET /api/v1/status | what the archive holds and what it was awake for |
| GET /data/record.db | refused on every plan: no key opens a copy of the archive. Bulk history is scoped one agreement at a time and delivered by us |
| GET /data/documents.ndjson.gz | by agreement only, not with a key: the documents corpus is the archive, and no plan opens it. Single documents stay public at /d/{mint}. |
curl -H "Authorization: Bearer $COT_KEY" https://chainoftitle.org/api/v1/token/<mint>
Tokens we have never seen
Coverage begins 2026-09-02 12:29 UTC. Ask for an older launch and we reconstruct it from the bonding curve's transaction history, up to a cap: thousands of archival RPC reads, which is a background job, not a request. A curve with more transactions than we read on request is refused and says so, rather than returning a partial record as a finished one. You get 202 with finding.class = "UNKNOWN" and a rebuild object; poll the same URL. A finished record is permanent, so the second call is usually the last one you ever make for that mint.
Reads are metered per key per UTC day, by plan. Rebuilds cost real money, so each caller can start 5 an hour and the service has a daily ceiling. When that is reached you get 503 rebuild_budget_exhausted: the archive is unaffected, only new reconstruction is paused. Bulk historical reconstruction is available under a data agreement: hello@chainoftitle.org.
Statuses
| 200 | a record. It may still be an UNKNOWN one. |
| 202 | accepted; a rebuild is queued or running. Retry-After is set. |
| 401 missing_key / invalid_key / revoked_key | no usable API key on the request |
| 429 quota_exceeded | the key's daily quota is used; it resets at 00:00 UTC |
| 400 not_an_address | not base58, or not 32–44 characters |
| 404 not_a_pump_launch | no pump.fun bonding curve exists for this address. A finding, not a failure. |
| 404 rebuild_failed | we tried to read the chain and could not. Our failure, not a finding. It says nothing about the token. |
| 429 rate_limited | too many rebuilds started from one address this hour |
| 503 rebuild_budget_exhausted / busy | we cannot pay for or keep up with more rebuilds right now |
Pay per read, no key
$0.01 a callThe two per-mint reads, /api/v1/token/{mint} and /api/v1/signals/{mint}, also answer over x402: a request without a key is offered the price, $0.01 in USDC on Solana, and a paid request is served with a receipt. The licence is per call, and one payment buys one read. Past roughly 1,000 reads a day, a key is cheaper.
The register as a tool (MCP)
freeAn MCP server at https://chainoftitle.org/mcp, no key and no account: add it to Claude or any MCP client as a remote connector and the agent can look a launch up by mint, search a ticker or name across the record, and read the archive's coverage figures. Answers are the free one-at-a-time surface with the citation URL attached - observations with their evidence, and a launch we did not observe answers as not observed. In Claude Code: claude mcp add --transport http chain-of-title https://chainoftitle.org/mcp.
Getting a key
$299 a month250,000 requests a day, the JSON launch record for any mint we hold, and reconstruction of launches we did not watch. Built for a wallet or a scanner asking about one launch at a time, as often as it needs to. Checkout is Stripe's; you then create your key in your account, where it is shown once and stored only as a hash. The record database as a file, the documents corpus and the creation feed are a data agreement instead, not a key.
Terms
| Cost | by plan, on the pricing page. Enterprise data agreements, an SLA, the live feed and bulk history are invoiced. |
| Licence | keyed responses are licensed for use in your own product and may not be redistributed in bulk. |
| CORS | open to every origin. Keep the key on your server: a key in browser code is readable by anyone. |
| Caching | a settled record is immutable and served max-age=3600, stale-while-revalidate=86400. Anything unsettled is no-store. |
| Volume | sustained use past the daily quota, or the creation feed itself, is an enterprise agreement - hello@chainoftitle.org. |
| Stability | fields are added, never repurposed. A breaking change gets a new version prefix and the old one keeps answering. |
| What it is not | not a price feed, not a signal, not advice. A clean record means none of the markers we record was present at launch, and says nothing about what it will do. |
If you ship this in front of users and find a record you think is wrong, tell us: a wrong entry on a record costs more than a missing one. hello@chainoftitle.org
What a response looks like
A real response from the current record - the clean-launch report's subject, refreshed as the record rebuilds. Keyed responses carry a license field beside this.
{
"apiVersion": "v1",
"coverage": {
"from": {
"ms": 1788352172581,
"iso": "2026-09-02T12:29:32.581Z"
},
"downtimeMinutes": 1844,
"note": "Launches outside these windows were not observed. We rebuild them from chain history on request; until that succeeds their provenance is unknown, not clean. downtimeMinutes is PROCESS downtime only - the minutes the collector was not running - and is not a measure of coverage: a feed that stays connected while dropping launches adds nothing to it. For coverage, read `launches` (observed at the creation transaction) against `records` (every row held); the difference was found later from chain by the scanner. See /corrections, coverage-downtime-understated."
},
"asOf": {
"ms": 1795358342108,
"iso": "2026-11-22T14:39:02.108Z"
},
"generatedAt": {
"ms": 1795366098066,
"iso": "2026-11-22T16:48:18.066Z"
},
"source": "https://chainoftitle.org/t/5NP3pTMgbvioJaki2Sdxm7DiaA2tt3BFkqTJyYNF5EwQ",
"license": "Chain of Title data licence (https://chainoftitle.org/terms)",
"mint": "5NP3pTMgbvioJaki2Sdxm7DiaA2tt3BFkqTJyYNF5EwQ",
"symbol": "PurpDrop",
"name": "PurpDrop",
"venue": "pumpfun",
"finding": {
"class": "NO_MARKERS",
"label": "Checked at birth, no markers found",
"why": "We watched this launch and ran every LAUNCH-RECORD test we have on it. None fired: the creator kept 1.7%, 428 outside wallets bought on the curve, the curve took longer than a minute to fill, and no single wallet took it. These tests read this launch's own first blocks and nothing else. What this launch shares with other launches - the same picture, the same document, the same creator - is counted separately, below, and is not part of this answer."
},
"cleanAtBirth": true,
"observedAtLaunch": true,
"origin": "observed",
"launch": {
"createdAt": {
"ms": 1788938126027,
"iso": "2026-09-09T07:15:26.027Z"
},
"creator": "DM5MVrMMr3vCacySqn5szpqDJMaPQCnSanoLAh7G7gLR",
"creatorSupplyPct": 1.737651816691,
"creatorSold": false,
"curveBuyers": 428,
"buyersFirst30s": 4,
"bundledBuyers": 0,
"graduated": true,
"graduationObserved": true,
"graduationCheck": null,
"graduationConfirmedBy": "pool",
"graduatedAt": {
"ms": 1788940589042,
"iso": "2026-09-09T07:56:29.042Z"
},
"secondsToGraduate": 2463,
"transaction": {
"signature": "4oMx1fn65LhFchEcDTPiY9Y3X3R74VVhkvnXHrGPosiXQmaic7pmVvVMkkw1bFKacK7cSfEg8z4C5ztXNfcxmetF",
"slot": 445551606
}
},
"curveBuyout": null,
"pool": null,
"funder": {
"sharedCohort": null
},
"shared": {
"document": {
"launches": 1,
"otherLaunches": 0,
"creators": 1,
"sha256": "76d15cc3504a4974dc315cb000ffb7714eaff5ee59537e7f595168191462e73e",
"launchesUrl": "https://chainoftitle.org/document/76d15cc3504a4974dc315cb000ffb7714eaff5ee59537e7f595168191462e73e"
},
"image": {
"launches": 1,
"otherLaunches": 0,
"creators": 1,
"sha256": "093f8bcf652cf53081e7bf366bff20448ca6dde2d3180e568d953e42ab12c75d",
"launchesUrl": "https://chainoftitle.org/picture/093f8bcf652cf53081e7bf366bff20448ca6dde2d3180e568d953e42ab12c75d"
},
"ticker": {
"launches": 1,
"creators": 1,
"note": "A ticker is not scarce. Many launches share one because the subject is popular, and this count alone distinguishes nothing."
},
"creator": {
"otherLaunches": 8,
"graduatedConfirmed": 0,
"launchesUrl": "https://chainoftitle.org/c/DM5MVrMMr3vCacySqn5szpqDJMaPQCnSanoLAh7G7gLR"
}
},
"flags": []
}