solana-launch-forensics 0.2.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- solana_launch_forensics-0.2.0/LICENSE +21 -0
- solana_launch_forensics-0.2.0/PKG-INFO +353 -0
- solana_launch_forensics-0.2.0/README.md +324 -0
- solana_launch_forensics-0.2.0/launchforensics/__init__.py +46 -0
- solana_launch_forensics-0.2.0/launchforensics/__main__.py +4 -0
- solana_launch_forensics-0.2.0/launchforensics/analytics.py +337 -0
- solana_launch_forensics-0.2.0/launchforensics/cli.py +437 -0
- solana_launch_forensics-0.2.0/launchforensics/cluster.py +157 -0
- solana_launch_forensics-0.2.0/launchforensics/dataset.py +74 -0
- solana_launch_forensics-0.2.0/launchforensics/pipeline.py +86 -0
- solana_launch_forensics-0.2.0/launchforensics/rpc.py +133 -0
- solana_launch_forensics-0.2.0/pyproject.toml +46 -0
- solana_launch_forensics-0.2.0/setup.cfg +4 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/PKG-INFO +353 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/SOURCES.txt +20 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/dependency_links.txt +1 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/entry_points.txt +2 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/requires.txt +3 -0
- solana_launch_forensics-0.2.0/solana_launch_forensics.egg-info/top_level.txt +1 -0
- solana_launch_forensics-0.2.0/tests/test_analytics.py +138 -0
- solana_launch_forensics-0.2.0/tests/test_cluster.py +268 -0
- solana_launch_forensics-0.2.0/tests/test_dataset.py +125 -0
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2026 solana-launch-forensics contributors
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
|
@@ -0,0 +1,353 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: solana-launch-forensics
|
|
3
|
+
Version: 0.2.0
|
|
4
|
+
Summary: Read-only forensics for pump.fun style token launches
|
|
5
|
+
Author: tdkrHuang
|
|
6
|
+
License: MIT
|
|
7
|
+
Project-URL: Homepage, https://github.com/tdkrHuang/solana-launch-forensics
|
|
8
|
+
Project-URL: Repository, https://github.com/tdkrHuang/solana-launch-forensics
|
|
9
|
+
Project-URL: Issues, https://github.com/tdkrHuang/solana-launch-forensics/issues
|
|
10
|
+
Project-URL: Changelog, https://github.com/tdkrHuang/solana-launch-forensics/releases
|
|
11
|
+
Keywords: solana,pump.fun,forensics,onchain,spl-token
|
|
12
|
+
Classifier: Development Status :: 4 - Beta
|
|
13
|
+
Classifier: Intended Audience :: Developers
|
|
14
|
+
Classifier: Programming Language :: Python :: 3
|
|
15
|
+
Classifier: Programming Language :: Python :: 3.9
|
|
16
|
+
Classifier: Programming Language :: Python :: 3.10
|
|
17
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
18
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
19
|
+
Classifier: Programming Language :: Python :: 3.13
|
|
20
|
+
Classifier: License :: OSI Approved :: MIT License
|
|
21
|
+
Classifier: Operating System :: OS Independent
|
|
22
|
+
Classifier: Topic :: Security
|
|
23
|
+
Requires-Python: >=3.9
|
|
24
|
+
Description-Content-Type: text/markdown
|
|
25
|
+
License-File: LICENSE
|
|
26
|
+
Provides-Extra: dev
|
|
27
|
+
Requires-Dist: pytest>=7; extra == "dev"
|
|
28
|
+
Dynamic: license-file
|
|
29
|
+
|
|
30
|
+
# solana-launch-forensics
|
|
31
|
+
|
|
32
|
+
Read-only forensics for pump.fun style token launches. **Standard library only** — no
|
|
33
|
+
API key, no `pip install` of crypto SDKs, no wallet connection.
|
|
34
|
+
|
|
35
|
+
It answers four questions from public chain data alone:
|
|
36
|
+
|
|
37
|
+
1. **Which mints did a wallet create?**
|
|
38
|
+
2. **How fast, and did any share a transaction?**
|
|
39
|
+
3. **How did the SOL leave, and did the payouts repeat one size?**
|
|
40
|
+
4. **Which mints belong to the same operator, and what is the evidence?**
|
|
41
|
+
|
|
42
|
+
Every output line is traceable to a signature you can paste into an explorer.
|
|
43
|
+
The tool never guesses intent.
|
|
44
|
+
|
|
45
|
+
```
|
|
46
|
+
$ launchforensics scan GtTg74Zp…4W14 --limit 60
|
|
47
|
+
wallet GtTg74Zp…4W14
|
|
48
|
+
scanned 60 transactions
|
|
49
|
+
creations 13
|
|
50
|
+
window 9m 11s
|
|
51
|
+
|
|
52
|
+
created (UTC) batch mint tx
|
|
53
|
+
2026-11-16 12:22:27 1 C53a9Fc6…pump ggoZiaudL3jx..6tiw
|
|
54
|
+
2026-11-16 12:23:07 1 CqAjDktv…pump 5BHGkpzjtjUU..RLY3
|
|
55
|
+
...
|
|
56
|
+
```
|
|
57
|
+
|
|
58
|
+
---
|
|
59
|
+
|
|
60
|
+
## Install
|
|
61
|
+
|
|
62
|
+
```bash
|
|
63
|
+
git clone https://github.com/tdkrHuang/solana-launch-forensics
|
|
64
|
+
cd solana-launch-forensics
|
|
65
|
+
python3 -m launchforensics --help
|
|
66
|
+
```
|
|
67
|
+
|
|
68
|
+
Python 3.9+. Nothing to install.
|
|
69
|
+
|
|
70
|
+
Optionally, as a command:
|
|
71
|
+
|
|
72
|
+
```bash
|
|
73
|
+
pip install -e .
|
|
74
|
+
launchforensics --help
|
|
75
|
+
```
|
|
76
|
+
|
|
77
|
+
---
|
|
78
|
+
|
|
79
|
+
## Commands
|
|
80
|
+
|
|
81
|
+
### `scan <wallet>` — mints created by a wallet
|
|
82
|
+
|
|
83
|
+
```bash
|
|
84
|
+
launchforensics scan <WALLET> --limit 400 --out wallet.json
|
|
85
|
+
```
|
|
86
|
+
|
|
87
|
+
| Flag | Meaning |
|
|
88
|
+
|---|---|
|
|
89
|
+
| `--limit N` | transactions to inspect (max 1000, newest first) |
|
|
90
|
+
| `--out FILE` | write the full JSON result |
|
|
91
|
+
| `--json` | print JSON to stdout |
|
|
92
|
+
|
|
93
|
+
Reports `creations`, the time `window` between the first and last creation, and any
|
|
94
|
+
transaction that created more than one mint.
|
|
95
|
+
|
|
96
|
+
### `tx <signature>` — one transaction, decoded
|
|
97
|
+
|
|
98
|
+
```bash
|
|
99
|
+
launchforensics tx <SIGNATURE>
|
|
100
|
+
```
|
|
101
|
+
|
|
102
|
+
Resolves which program emitted each `Program log:` line by walking the
|
|
103
|
+
`invoke` / `success` stack, so pump.fun operations are counted separately from
|
|
104
|
+
token-program noise. Output includes signers, fee, pump instruction counts,
|
|
105
|
+
referenced mints, SOL transfers and repeated equal-size outflows.
|
|
106
|
+
|
|
107
|
+
### `flows <wallet>` — repeated equal-size payouts
|
|
108
|
+
|
|
109
|
+
```bash
|
|
110
|
+
launchforensics flows <WALLET> --limit 200 --min-sol 0.01 --across-tx
|
|
111
|
+
```
|
|
112
|
+
|
|
113
|
+
Groups SOL transfers leaving the wallet by exact lamport amount. Scripted splitters
|
|
114
|
+
tend to pay one number to many destinations; humans rarely do.
|
|
115
|
+
|
|
116
|
+
| Flag | Meaning |
|
|
117
|
+
|---|---|
|
|
118
|
+
| `--across-tx` | group identical amounts **across** transactions (default: within one tx) |
|
|
119
|
+
| `--min-count N` | minimum distinct destinations to report |
|
|
120
|
+
| `--min-sol X` | ignore dust below X SOL |
|
|
121
|
+
|
|
122
|
+
Real output — nine wallets received the identical 486,609,148 lamports:
|
|
123
|
+
|
|
124
|
+
```
|
|
125
|
+
$ launchforensics flows GtTg74Zp… --limit 150 --min-sol 0.4 --across-tx
|
|
126
|
+
0.486609148 SOL x 9 destinations (9 txs)
|
|
127
|
+
-> 2c5o9kpr…vJq4
|
|
128
|
+
-> 3EyicXms…hPb
|
|
129
|
+
-> 6NmwtSZ5…M2N
|
|
130
|
+
...
|
|
131
|
+
```
|
|
132
|
+
|
|
133
|
+
`--across-tx` is the mode that matters for payout analysis: a splitter usually pays
|
|
134
|
+
one destination per transaction, so a within-transaction view finds nothing.
|
|
135
|
+
|
|
136
|
+
### `cluster <wallet>` — mints that share evidence
|
|
137
|
+
|
|
138
|
+
```bash
|
|
139
|
+
launchforensics cluster <WALLET> --limit 400
|
|
140
|
+
```
|
|
141
|
+
|
|
142
|
+
Builds a graph whose nodes are wallets and mints and whose edges are individual,
|
|
143
|
+
signed facts:
|
|
144
|
+
|
|
145
|
+
| Edge | Meaning |
|
|
146
|
+
|---|---|
|
|
147
|
+
| `created` | the wallet paid for the transaction that created the mint |
|
|
148
|
+
| `swept` | the wallet collected that mint's creator fee — only the creator can |
|
|
149
|
+
| `funded` | SOL moved from one wallet to another |
|
|
150
|
+
|
|
151
|
+
Connected components containing at least two mints are printed together, with the
|
|
152
|
+
signature of every link, so any single claim can be checked on an explorer.
|
|
153
|
+
|
|
154
|
+
| Flag | Meaning |
|
|
155
|
+
|---|---|
|
|
156
|
+
| `--min-mints N` | ignore components below N mints (default 2) |
|
|
157
|
+
| `--max-mints N` | mints printed per cluster (default 25) |
|
|
158
|
+
| `--max-edges N` | evidence lines printed per cluster (default 25) |
|
|
159
|
+
| `--out FILE` | write the full JSON result |
|
|
160
|
+
| `--json` | print JSON to stdout |
|
|
161
|
+
|
|
162
|
+
Example output (addresses shortened):
|
|
163
|
+
|
|
164
|
+
```
|
|
165
|
+
wallet GtTg74ZpUb…4W14
|
|
166
|
+
scanned 5 transactions
|
|
167
|
+
evidence created 3, funded 1, swept 3
|
|
168
|
+
clusters 1 (minimum 2 mints)
|
|
169
|
+
|
|
170
|
+
cluster 1: 3 mints, 2 wallets
|
|
171
|
+
links created 3, funded 1, swept 3
|
|
172
|
+
window 2025-11-16 04:50:00 -> 2025-11-16 05:16:21 (26m 21s)
|
|
173
|
+
wallets
|
|
174
|
+
- GtTg74ZpUb…4W14
|
|
175
|
+
- 8fJ1XKz6Yb…QwZa
|
|
176
|
+
mints
|
|
177
|
+
- EGX8MJT8wd…pump
|
|
178
|
+
- 2ZZhunwtYT…pump
|
|
179
|
+
- 2GoDADTFmh…pump
|
|
180
|
+
evidence
|
|
181
|
+
created GtTg74ZpUb…4W14 -> EGX8MJT8wd…pump create1
|
|
182
|
+
funded 8fJ1XKz6Yb…QwZa -> GtTg74ZpUb…4W14 0.05 SOL fund1
|
|
183
|
+
swept GtTg74ZpUb…4W14 -> EGX8MJT8wd…pump sweep1
|
|
184
|
+
```
|
|
185
|
+
|
|
186
|
+
A control wallet — one that created a single token and collected nothing — prints
|
|
187
|
+
`No cluster above the threshold.` That silence is the point: the command reports
|
|
188
|
+
links it can sign, not wallets it finds interesting.
|
|
189
|
+
|
|
190
|
+
Shared funding is a link, not proof. Exchanges and faucets fund thousands of
|
|
191
|
+
unrelated wallets, so read the `funded` edges with that in mind; `created` and
|
|
192
|
+
`swept` are the stronger links because both require control of the wallet.
|
|
193
|
+
|
|
194
|
+
---
|
|
195
|
+
|
|
196
|
+
## Dataset
|
|
197
|
+
|
|
198
|
+
`datasets/clusters.jsonl` collects reproducible cluster records: the seed wallet,
|
|
199
|
+
the exact command, the scan bound, the evidence counts and a confidence grade.
|
|
200
|
+
The schema is documented in [`datasets/SCHEMA.md`](datasets/SCHEMA.md).
|
|
201
|
+
|
|
202
|
+
```bash
|
|
203
|
+
python3 tools/build_dataset.py --seeds datasets/seeds.json --limit 200
|
|
204
|
+
python3 tools/build_dataset.py --seeds datasets/seeds.json --dry-run
|
|
205
|
+
```
|
|
206
|
+
|
|
207
|
+
Grading is by the weakest link used to join the mints:
|
|
208
|
+
|
|
209
|
+
| Grade | Condition |
|
|
210
|
+
|---|---|
|
|
211
|
+
| `high` | at least one `swept` edge — only a creator can collect the fee |
|
|
212
|
+
| `medium` | two or more `created` edges from one wallet |
|
|
213
|
+
| `low` | joined only by funding transfers (dropped by default) |
|
|
214
|
+
|
|
215
|
+
Records are keyed by a hash of their mints, so re-running updates instead of
|
|
216
|
+
duplicating, and a rate-limited run keeps what it already wrote.
|
|
217
|
+
|
|
218
|
+
---
|
|
219
|
+
|
|
220
|
+
## What it correctly distinguishes
|
|
221
|
+
|
|
222
|
+
Two traps that produce wrong conclusions:
|
|
223
|
+
|
|
224
|
+
**1. `Program log:` lines carry no program id.**
|
|
225
|
+
|
|
226
|
+
A pump.fun `CreateV2` fans out into Token-2022, which logs its own
|
|
227
|
+
`Instruction: InitializeMint2`. Naive substring counting reports two creations per
|
|
228
|
+
launch. This tool attributes logs through the invoke stack, so `create_count`
|
|
229
|
+
returns `1`.
|
|
230
|
+
|
|
231
|
+
**2. Touching a mint is not creating it.**
|
|
232
|
+
|
|
233
|
+
A creator-fee sweep references seven mints and contains zero creations:
|
|
234
|
+
|
|
235
|
+
```
|
|
236
|
+
$ launchforensics tx zWcwg1pBsstXjKUw…whEq
|
|
237
|
+
pump ops SweepCreatorFee x7, CollectCreatorFeeV2
|
|
238
|
+
creates 0
|
|
239
|
+
mints 7 referenced
|
|
240
|
+
```
|
|
241
|
+
|
|
242
|
+
That distinction matters: a sweep across seven mints proves one wallet controls the
|
|
243
|
+
creator-fee accounts of all seven. Calling it "seven launches in one transaction"
|
|
244
|
+
would be wrong, and a reviewer will check.
|
|
245
|
+
|
|
246
|
+
---
|
|
247
|
+
|
|
248
|
+
## Output schema
|
|
249
|
+
|
|
250
|
+
`scan` writes:
|
|
251
|
+
|
|
252
|
+
```json
|
|
253
|
+
{
|
|
254
|
+
"wallet": "...",
|
|
255
|
+
"scanned_transactions": 400,
|
|
256
|
+
"creation_count": 27,
|
|
257
|
+
"window_seconds": 2350,
|
|
258
|
+
"creations": [
|
|
259
|
+
{
|
|
260
|
+
"mint": "...pump",
|
|
261
|
+
"signature": "...",
|
|
262
|
+
"slot": 470014561,
|
|
263
|
+
"block_time": 1763268569,
|
|
264
|
+
"batch_size": 1
|
|
265
|
+
}
|
|
266
|
+
],
|
|
267
|
+
"batch_transactions": []
|
|
268
|
+
}
|
|
269
|
+
```
|
|
270
|
+
|
|
271
|
+
`cluster` writes:
|
|
272
|
+
|
|
273
|
+
```json
|
|
274
|
+
{
|
|
275
|
+
"wallet": "...",
|
|
276
|
+
"scanned_transactions": 400,
|
|
277
|
+
"edge_counts": {"created": 84, "funded": 12, "swept": 61},
|
|
278
|
+
"cluster_count": 2,
|
|
279
|
+
"clusters": [
|
|
280
|
+
{
|
|
281
|
+
"mints": ["...pump", "...pump"],
|
|
282
|
+
"wallets": ["..."],
|
|
283
|
+
"links": {"created": 84, "funded": 12, "swept": 61},
|
|
284
|
+
"first_block_time": 1763268569,
|
|
285
|
+
"last_block_time": 1763272285,
|
|
286
|
+
"edges": [
|
|
287
|
+
{"type": "created", "wallet": "...", "mint": "...pump", "signature": "..."}
|
|
288
|
+
]
|
|
289
|
+
}
|
|
290
|
+
]
|
|
291
|
+
}
|
|
292
|
+
```
|
|
293
|
+
|
|
294
|
+
---
|
|
295
|
+
|
|
296
|
+
## Limits, stated plainly
|
|
297
|
+
|
|
298
|
+
- **Public RPC is rate limited.** Default pause is 0.35 s and requests retry with
|
|
299
|
+
backoff. `scan` skips individual failed transactions and reports
|
|
300
|
+
`failed_transactions` instead of aborting. If you see many failures, switch
|
|
301
|
+
endpoint:
|
|
302
|
+
|
|
303
|
+
```bash
|
|
304
|
+
launchforensics --rpc https://solana-rpc.publicnode.com scan <WALLET> --limit 400
|
|
305
|
+
```
|
|
306
|
+
|
|
307
|
+
For serious volume, point `--rpc` (or `SOLANA_RPC_URL`) at your own node.
|
|
308
|
+
- **`--limit` caps history.** `scan` reads newest-first; a wallet older than the
|
|
309
|
+
window will appear to have fewer creations. Results say `scanned_transactions`
|
|
310
|
+
so the bound is visible.
|
|
311
|
+
- **pump.fun mints are detected by the `pump` suffix.** Non-pump SPL mints are only
|
|
312
|
+
found when the RPC returns parsed `initializeMint` instructions.
|
|
313
|
+
- **No price or PnL data.** Deliberately. Prices come from third-party APIs with
|
|
314
|
+
their own reliability; this tool only reports what the chain itself records.
|
|
315
|
+
- **Endpoints disagree.** Some public RPCs return an empty signature list for
|
|
316
|
+
addresses they do not index — including wallets that are active. `scan` will
|
|
317
|
+
report `no signatures returned`. Confirm on a second endpoint before treating
|
|
318
|
+
a wallet as inactive, and say which endpoint you used.
|
|
319
|
+
- **Not a verdict.** Repeated equal-size payouts and batched creations are patterns,
|
|
320
|
+
not proof of wrongdoing.
|
|
321
|
+
|
|
322
|
+
---
|
|
323
|
+
|
|
324
|
+
## Paid reports
|
|
325
|
+
|
|
326
|
+
Reading the code is free. If you need this applied to a specific wallet — with
|
|
327
|
+
the funding chain, payout reconstruction and a PDF you can hand to a community,
|
|
328
|
+
a lawyer or an exchange — I write those reports with this tool:
|
|
329
|
+
|
|
330
|
+
- **Lite $99** — one wallet, 48h
|
|
331
|
+
- **Standard $299** — up to 5 wallets, 72h
|
|
332
|
+
- **Deep $699** — up to 15 wallets, legal appendix, 5 business days
|
|
333
|
+
|
|
334
|
+
Every claim carries a signature. Sample report (real case, 96 launches):
|
|
335
|
+
[docs/sample-report.md](docs/sample-report.md)
|
|
336
|
+
|
|
337
|
+
## Tests
|
|
338
|
+
|
|
339
|
+
```bash
|
|
340
|
+
python3 -m pytest -q
|
|
341
|
+
```
|
|
342
|
+
|
|
343
|
+
Fixtures in `tests/fixtures.py` mirror real responses, including a versioned
|
|
344
|
+
transaction **without `message.header`** and a Token-2022 `InitializeMint2` nested
|
|
345
|
+
inside a pump.fun call. `tests/test_cluster.py` covers the graph rules: a single
|
|
346
|
+
mint is never a cluster, shared funding joins two wallets, a sweep joins mints to
|
|
347
|
+
the sweeper, and duplicate links collapse without duplicating mints.
|
|
348
|
+
|
|
349
|
+
---
|
|
350
|
+
|
|
351
|
+
## License
|
|
352
|
+
|
|
353
|
+
MIT
|
|
@@ -0,0 +1,324 @@
|
|
|
1
|
+
# solana-launch-forensics
|
|
2
|
+
|
|
3
|
+
Read-only forensics for pump.fun style token launches. **Standard library only** — no
|
|
4
|
+
API key, no `pip install` of crypto SDKs, no wallet connection.
|
|
5
|
+
|
|
6
|
+
It answers four questions from public chain data alone:
|
|
7
|
+
|
|
8
|
+
1. **Which mints did a wallet create?**
|
|
9
|
+
2. **How fast, and did any share a transaction?**
|
|
10
|
+
3. **How did the SOL leave, and did the payouts repeat one size?**
|
|
11
|
+
4. **Which mints belong to the same operator, and what is the evidence?**
|
|
12
|
+
|
|
13
|
+
Every output line is traceable to a signature you can paste into an explorer.
|
|
14
|
+
The tool never guesses intent.
|
|
15
|
+
|
|
16
|
+
```
|
|
17
|
+
$ launchforensics scan GtTg74Zp…4W14 --limit 60
|
|
18
|
+
wallet GtTg74Zp…4W14
|
|
19
|
+
scanned 60 transactions
|
|
20
|
+
creations 13
|
|
21
|
+
window 9m 11s
|
|
22
|
+
|
|
23
|
+
created (UTC) batch mint tx
|
|
24
|
+
2026-11-16 12:22:27 1 C53a9Fc6…pump ggoZiaudL3jx..6tiw
|
|
25
|
+
2026-11-16 12:23:07 1 CqAjDktv…pump 5BHGkpzjtjUU..RLY3
|
|
26
|
+
...
|
|
27
|
+
```
|
|
28
|
+
|
|
29
|
+
---
|
|
30
|
+
|
|
31
|
+
## Install
|
|
32
|
+
|
|
33
|
+
```bash
|
|
34
|
+
git clone https://github.com/tdkrHuang/solana-launch-forensics
|
|
35
|
+
cd solana-launch-forensics
|
|
36
|
+
python3 -m launchforensics --help
|
|
37
|
+
```
|
|
38
|
+
|
|
39
|
+
Python 3.9+. Nothing to install.
|
|
40
|
+
|
|
41
|
+
Optionally, as a command:
|
|
42
|
+
|
|
43
|
+
```bash
|
|
44
|
+
pip install -e .
|
|
45
|
+
launchforensics --help
|
|
46
|
+
```
|
|
47
|
+
|
|
48
|
+
---
|
|
49
|
+
|
|
50
|
+
## Commands
|
|
51
|
+
|
|
52
|
+
### `scan <wallet>` — mints created by a wallet
|
|
53
|
+
|
|
54
|
+
```bash
|
|
55
|
+
launchforensics scan <WALLET> --limit 400 --out wallet.json
|
|
56
|
+
```
|
|
57
|
+
|
|
58
|
+
| Flag | Meaning |
|
|
59
|
+
|---|---|
|
|
60
|
+
| `--limit N` | transactions to inspect (max 1000, newest first) |
|
|
61
|
+
| `--out FILE` | write the full JSON result |
|
|
62
|
+
| `--json` | print JSON to stdout |
|
|
63
|
+
|
|
64
|
+
Reports `creations`, the time `window` between the first and last creation, and any
|
|
65
|
+
transaction that created more than one mint.
|
|
66
|
+
|
|
67
|
+
### `tx <signature>` — one transaction, decoded
|
|
68
|
+
|
|
69
|
+
```bash
|
|
70
|
+
launchforensics tx <SIGNATURE>
|
|
71
|
+
```
|
|
72
|
+
|
|
73
|
+
Resolves which program emitted each `Program log:` line by walking the
|
|
74
|
+
`invoke` / `success` stack, so pump.fun operations are counted separately from
|
|
75
|
+
token-program noise. Output includes signers, fee, pump instruction counts,
|
|
76
|
+
referenced mints, SOL transfers and repeated equal-size outflows.
|
|
77
|
+
|
|
78
|
+
### `flows <wallet>` — repeated equal-size payouts
|
|
79
|
+
|
|
80
|
+
```bash
|
|
81
|
+
launchforensics flows <WALLET> --limit 200 --min-sol 0.01 --across-tx
|
|
82
|
+
```
|
|
83
|
+
|
|
84
|
+
Groups SOL transfers leaving the wallet by exact lamport amount. Scripted splitters
|
|
85
|
+
tend to pay one number to many destinations; humans rarely do.
|
|
86
|
+
|
|
87
|
+
| Flag | Meaning |
|
|
88
|
+
|---|---|
|
|
89
|
+
| `--across-tx` | group identical amounts **across** transactions (default: within one tx) |
|
|
90
|
+
| `--min-count N` | minimum distinct destinations to report |
|
|
91
|
+
| `--min-sol X` | ignore dust below X SOL |
|
|
92
|
+
|
|
93
|
+
Real output — nine wallets received the identical 486,609,148 lamports:
|
|
94
|
+
|
|
95
|
+
```
|
|
96
|
+
$ launchforensics flows GtTg74Zp… --limit 150 --min-sol 0.4 --across-tx
|
|
97
|
+
0.486609148 SOL x 9 destinations (9 txs)
|
|
98
|
+
-> 2c5o9kpr…vJq4
|
|
99
|
+
-> 3EyicXms…hPb
|
|
100
|
+
-> 6NmwtSZ5…M2N
|
|
101
|
+
...
|
|
102
|
+
```
|
|
103
|
+
|
|
104
|
+
`--across-tx` is the mode that matters for payout analysis: a splitter usually pays
|
|
105
|
+
one destination per transaction, so a within-transaction view finds nothing.
|
|
106
|
+
|
|
107
|
+
### `cluster <wallet>` — mints that share evidence
|
|
108
|
+
|
|
109
|
+
```bash
|
|
110
|
+
launchforensics cluster <WALLET> --limit 400
|
|
111
|
+
```
|
|
112
|
+
|
|
113
|
+
Builds a graph whose nodes are wallets and mints and whose edges are individual,
|
|
114
|
+
signed facts:
|
|
115
|
+
|
|
116
|
+
| Edge | Meaning |
|
|
117
|
+
|---|---|
|
|
118
|
+
| `created` | the wallet paid for the transaction that created the mint |
|
|
119
|
+
| `swept` | the wallet collected that mint's creator fee — only the creator can |
|
|
120
|
+
| `funded` | SOL moved from one wallet to another |
|
|
121
|
+
|
|
122
|
+
Connected components containing at least two mints are printed together, with the
|
|
123
|
+
signature of every link, so any single claim can be checked on an explorer.
|
|
124
|
+
|
|
125
|
+
| Flag | Meaning |
|
|
126
|
+
|---|---|
|
|
127
|
+
| `--min-mints N` | ignore components below N mints (default 2) |
|
|
128
|
+
| `--max-mints N` | mints printed per cluster (default 25) |
|
|
129
|
+
| `--max-edges N` | evidence lines printed per cluster (default 25) |
|
|
130
|
+
| `--out FILE` | write the full JSON result |
|
|
131
|
+
| `--json` | print JSON to stdout |
|
|
132
|
+
|
|
133
|
+
Example output (addresses shortened):
|
|
134
|
+
|
|
135
|
+
```
|
|
136
|
+
wallet GtTg74ZpUb…4W14
|
|
137
|
+
scanned 5 transactions
|
|
138
|
+
evidence created 3, funded 1, swept 3
|
|
139
|
+
clusters 1 (minimum 2 mints)
|
|
140
|
+
|
|
141
|
+
cluster 1: 3 mints, 2 wallets
|
|
142
|
+
links created 3, funded 1, swept 3
|
|
143
|
+
window 2025-11-16 04:50:00 -> 2025-11-16 05:16:21 (26m 21s)
|
|
144
|
+
wallets
|
|
145
|
+
- GtTg74ZpUb…4W14
|
|
146
|
+
- 8fJ1XKz6Yb…QwZa
|
|
147
|
+
mints
|
|
148
|
+
- EGX8MJT8wd…pump
|
|
149
|
+
- 2ZZhunwtYT…pump
|
|
150
|
+
- 2GoDADTFmh…pump
|
|
151
|
+
evidence
|
|
152
|
+
created GtTg74ZpUb…4W14 -> EGX8MJT8wd…pump create1
|
|
153
|
+
funded 8fJ1XKz6Yb…QwZa -> GtTg74ZpUb…4W14 0.05 SOL fund1
|
|
154
|
+
swept GtTg74ZpUb…4W14 -> EGX8MJT8wd…pump sweep1
|
|
155
|
+
```
|
|
156
|
+
|
|
157
|
+
A control wallet — one that created a single token and collected nothing — prints
|
|
158
|
+
`No cluster above the threshold.` That silence is the point: the command reports
|
|
159
|
+
links it can sign, not wallets it finds interesting.
|
|
160
|
+
|
|
161
|
+
Shared funding is a link, not proof. Exchanges and faucets fund thousands of
|
|
162
|
+
unrelated wallets, so read the `funded` edges with that in mind; `created` and
|
|
163
|
+
`swept` are the stronger links because both require control of the wallet.
|
|
164
|
+
|
|
165
|
+
---
|
|
166
|
+
|
|
167
|
+
## Dataset
|
|
168
|
+
|
|
169
|
+
`datasets/clusters.jsonl` collects reproducible cluster records: the seed wallet,
|
|
170
|
+
the exact command, the scan bound, the evidence counts and a confidence grade.
|
|
171
|
+
The schema is documented in [`datasets/SCHEMA.md`](datasets/SCHEMA.md).
|
|
172
|
+
|
|
173
|
+
```bash
|
|
174
|
+
python3 tools/build_dataset.py --seeds datasets/seeds.json --limit 200
|
|
175
|
+
python3 tools/build_dataset.py --seeds datasets/seeds.json --dry-run
|
|
176
|
+
```
|
|
177
|
+
|
|
178
|
+
Grading is by the weakest link used to join the mints:
|
|
179
|
+
|
|
180
|
+
| Grade | Condition |
|
|
181
|
+
|---|---|
|
|
182
|
+
| `high` | at least one `swept` edge — only a creator can collect the fee |
|
|
183
|
+
| `medium` | two or more `created` edges from one wallet |
|
|
184
|
+
| `low` | joined only by funding transfers (dropped by default) |
|
|
185
|
+
|
|
186
|
+
Records are keyed by a hash of their mints, so re-running updates instead of
|
|
187
|
+
duplicating, and a rate-limited run keeps what it already wrote.
|
|
188
|
+
|
|
189
|
+
---
|
|
190
|
+
|
|
191
|
+
## What it correctly distinguishes
|
|
192
|
+
|
|
193
|
+
Two traps that produce wrong conclusions:
|
|
194
|
+
|
|
195
|
+
**1. `Program log:` lines carry no program id.**
|
|
196
|
+
|
|
197
|
+
A pump.fun `CreateV2` fans out into Token-2022, which logs its own
|
|
198
|
+
`Instruction: InitializeMint2`. Naive substring counting reports two creations per
|
|
199
|
+
launch. This tool attributes logs through the invoke stack, so `create_count`
|
|
200
|
+
returns `1`.
|
|
201
|
+
|
|
202
|
+
**2. Touching a mint is not creating it.**
|
|
203
|
+
|
|
204
|
+
A creator-fee sweep references seven mints and contains zero creations:
|
|
205
|
+
|
|
206
|
+
```
|
|
207
|
+
$ launchforensics tx zWcwg1pBsstXjKUw…whEq
|
|
208
|
+
pump ops SweepCreatorFee x7, CollectCreatorFeeV2
|
|
209
|
+
creates 0
|
|
210
|
+
mints 7 referenced
|
|
211
|
+
```
|
|
212
|
+
|
|
213
|
+
That distinction matters: a sweep across seven mints proves one wallet controls the
|
|
214
|
+
creator-fee accounts of all seven. Calling it "seven launches in one transaction"
|
|
215
|
+
would be wrong, and a reviewer will check.
|
|
216
|
+
|
|
217
|
+
---
|
|
218
|
+
|
|
219
|
+
## Output schema
|
|
220
|
+
|
|
221
|
+
`scan` writes:
|
|
222
|
+
|
|
223
|
+
```json
|
|
224
|
+
{
|
|
225
|
+
"wallet": "...",
|
|
226
|
+
"scanned_transactions": 400,
|
|
227
|
+
"creation_count": 27,
|
|
228
|
+
"window_seconds": 2350,
|
|
229
|
+
"creations": [
|
|
230
|
+
{
|
|
231
|
+
"mint": "...pump",
|
|
232
|
+
"signature": "...",
|
|
233
|
+
"slot": 470014561,
|
|
234
|
+
"block_time": 1763268569,
|
|
235
|
+
"batch_size": 1
|
|
236
|
+
}
|
|
237
|
+
],
|
|
238
|
+
"batch_transactions": []
|
|
239
|
+
}
|
|
240
|
+
```
|
|
241
|
+
|
|
242
|
+
`cluster` writes:
|
|
243
|
+
|
|
244
|
+
```json
|
|
245
|
+
{
|
|
246
|
+
"wallet": "...",
|
|
247
|
+
"scanned_transactions": 400,
|
|
248
|
+
"edge_counts": {"created": 84, "funded": 12, "swept": 61},
|
|
249
|
+
"cluster_count": 2,
|
|
250
|
+
"clusters": [
|
|
251
|
+
{
|
|
252
|
+
"mints": ["...pump", "...pump"],
|
|
253
|
+
"wallets": ["..."],
|
|
254
|
+
"links": {"created": 84, "funded": 12, "swept": 61},
|
|
255
|
+
"first_block_time": 1763268569,
|
|
256
|
+
"last_block_time": 1763272285,
|
|
257
|
+
"edges": [
|
|
258
|
+
{"type": "created", "wallet": "...", "mint": "...pump", "signature": "..."}
|
|
259
|
+
]
|
|
260
|
+
}
|
|
261
|
+
]
|
|
262
|
+
}
|
|
263
|
+
```
|
|
264
|
+
|
|
265
|
+
---
|
|
266
|
+
|
|
267
|
+
## Limits, stated plainly
|
|
268
|
+
|
|
269
|
+
- **Public RPC is rate limited.** Default pause is 0.35 s and requests retry with
|
|
270
|
+
backoff. `scan` skips individual failed transactions and reports
|
|
271
|
+
`failed_transactions` instead of aborting. If you see many failures, switch
|
|
272
|
+
endpoint:
|
|
273
|
+
|
|
274
|
+
```bash
|
|
275
|
+
launchforensics --rpc https://solana-rpc.publicnode.com scan <WALLET> --limit 400
|
|
276
|
+
```
|
|
277
|
+
|
|
278
|
+
For serious volume, point `--rpc` (or `SOLANA_RPC_URL`) at your own node.
|
|
279
|
+
- **`--limit` caps history.** `scan` reads newest-first; a wallet older than the
|
|
280
|
+
window will appear to have fewer creations. Results say `scanned_transactions`
|
|
281
|
+
so the bound is visible.
|
|
282
|
+
- **pump.fun mints are detected by the `pump` suffix.** Non-pump SPL mints are only
|
|
283
|
+
found when the RPC returns parsed `initializeMint` instructions.
|
|
284
|
+
- **No price or PnL data.** Deliberately. Prices come from third-party APIs with
|
|
285
|
+
their own reliability; this tool only reports what the chain itself records.
|
|
286
|
+
- **Endpoints disagree.** Some public RPCs return an empty signature list for
|
|
287
|
+
addresses they do not index — including wallets that are active. `scan` will
|
|
288
|
+
report `no signatures returned`. Confirm on a second endpoint before treating
|
|
289
|
+
a wallet as inactive, and say which endpoint you used.
|
|
290
|
+
- **Not a verdict.** Repeated equal-size payouts and batched creations are patterns,
|
|
291
|
+
not proof of wrongdoing.
|
|
292
|
+
|
|
293
|
+
---
|
|
294
|
+
|
|
295
|
+
## Paid reports
|
|
296
|
+
|
|
297
|
+
Reading the code is free. If you need this applied to a specific wallet — with
|
|
298
|
+
the funding chain, payout reconstruction and a PDF you can hand to a community,
|
|
299
|
+
a lawyer or an exchange — I write those reports with this tool:
|
|
300
|
+
|
|
301
|
+
- **Lite $99** — one wallet, 48h
|
|
302
|
+
- **Standard $299** — up to 5 wallets, 72h
|
|
303
|
+
- **Deep $699** — up to 15 wallets, legal appendix, 5 business days
|
|
304
|
+
|
|
305
|
+
Every claim carries a signature. Sample report (real case, 96 launches):
|
|
306
|
+
[docs/sample-report.md](docs/sample-report.md)
|
|
307
|
+
|
|
308
|
+
## Tests
|
|
309
|
+
|
|
310
|
+
```bash
|
|
311
|
+
python3 -m pytest -q
|
|
312
|
+
```
|
|
313
|
+
|
|
314
|
+
Fixtures in `tests/fixtures.py` mirror real responses, including a versioned
|
|
315
|
+
transaction **without `message.header`** and a Token-2022 `InitializeMint2` nested
|
|
316
|
+
inside a pump.fun call. `tests/test_cluster.py` covers the graph rules: a single
|
|
317
|
+
mint is never a cluster, shared funding joins two wallets, a sweep joins mints to
|
|
318
|
+
the sweeper, and duplicate links collapse without duplicating mints.
|
|
319
|
+
|
|
320
|
+
---
|
|
321
|
+
|
|
322
|
+
## License
|
|
323
|
+
|
|
324
|
+
MIT
|