<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: tdkrHuang</title>
    <description>The latest articles on DEV Community by tdkrHuang (@tdkrhuang).</description>
    <link>https://dev.to/tdkrhuang</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4251964%2F71b69ce8-6273-4a8d-aebf-084d121a3cfd.png</url>
      <title>DEV Community: tdkrHuang</title>
      <link>https://dev.to/tdkrhuang</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/tdkrhuang"/>
    <language>en</language>
    <item>
      <title>Reading Solana launch logs correctly: a fee sweep is not a creation</title>
      <dc:creator>tdkrHuang</dc:creator>
      <pubDate>Wed, 18 Nov 2026 13:18:19 +0000</pubDate>
      <link>https://dev.to/tdkrhuang/reading-solana-launch-logs-correctly-a-fee-sweep-is-not-a-creation-4gf4</link>
      <guid>https://dev.to/tdkrhuang/reading-solana-launch-logs-correctly-a-fee-sweep-is-not-a-creation-4gf4</guid>
      <description>&lt;p&gt;I spent a weekend tracing token launches on Solana, and the most useful thing I&lt;br&gt;
learned was not about clustering wallets. It was that the logs are easy to read&lt;br&gt;
in a way that produces the opposite of the truth.&lt;/p&gt;

&lt;p&gt;Here is the transaction that taught me.&lt;/p&gt;
&lt;h2&gt;
  
  
  What the log looked like
&lt;/h2&gt;

&lt;p&gt;A single transaction contained seven pump.fun instructions. Counting them gave&lt;br&gt;
an obvious headline: &lt;em&gt;seven tokens created in one transaction&lt;/em&gt;. That claim&lt;br&gt;
circulated in a draft before I checked it properly.&lt;/p&gt;

&lt;p&gt;It was wrong. Those seven instructions were &lt;code&gt;SweepCreatorFee&lt;/code&gt;, not &lt;code&gt;CreateV2&lt;/code&gt;.&lt;br&gt;
The transaction swept the creator fees of seven &lt;em&gt;existing&lt;/em&gt; tokens. Zero tokens&lt;br&gt;
were created in it.&lt;/p&gt;

&lt;p&gt;The difference matters because creator fees can only be collected by the token's&lt;br&gt;
creator. So the correct claim is not weaker than the wrong one — it is stronger.&lt;br&gt;
Seven tokens swept in one transaction means one wallet controls the creator-fee&lt;br&gt;
rights of all seven. That is not a behavioural inference. It is the ownership&lt;br&gt;
record itself.&lt;/p&gt;
&lt;h2&gt;
  
  
  Why the naive reading fails
&lt;/h2&gt;

&lt;p&gt;Two traps, and they compound.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. Inner instructions are nested.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;A pump.fun &lt;code&gt;CreateV2&lt;/code&gt; calls Token-2022, which logs its own&lt;br&gt;
&lt;code&gt;Instruction: InitializeMint2&lt;/code&gt;. If you count &lt;code&gt;InitializeMint2&lt;/code&gt; occurrences you&lt;br&gt;
report two creations per launch. If you count pump.fun instructions you miss&lt;br&gt;
that the nested call happened at all.&lt;/p&gt;

&lt;p&gt;The fix is to reconstruct the &lt;code&gt;invoke&lt;/code&gt; / &lt;code&gt;success&lt;/code&gt; stack from the program logs&lt;br&gt;
and attribute every inner instruction to the program that actually invoked it.&lt;br&gt;
Positional matching breaks as soon as calls nest more than one level deep.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. &lt;code&gt;Program log:&lt;/code&gt; lines carry no program id.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Solana's log stream interleaves messages from every program in the call chain.&lt;br&gt;
A line reading &lt;code&gt;Program log: Instruction: InitializeMint2&lt;/code&gt; tells you that &lt;em&gt;some&lt;/em&gt;&lt;br&gt;
program emitted it. Without the invoke stack, you cannot say which.&lt;/p&gt;
&lt;h2&gt;
  
  
  What the correct output looks like
&lt;/h2&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;$ launchforensics tx &amp;lt;signature&amp;gt;

pump ops    SweepCreatorFee x7, CollectCreatorFeeV2
creates     0
mints       7 referenced
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;Compare that with a real creation:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;$ launchforensics scan &amp;lt;wallet&amp;gt; --limit 400

wallet          ...
scanned         400 transactions
creations       27
window          39m 11s
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Both are readable in a terminal, and every line carries the signature that&lt;br&gt;
produced it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The pattern that only shows up across transactions
&lt;/h2&gt;

&lt;p&gt;One more thing worth checking, because a single-transaction view returns zero.&lt;/p&gt;

&lt;p&gt;A splitter usually pays one destination per transaction. So if you want to know&lt;br&gt;
whether payouts repeat an exact size, you have to look across signatures and&lt;br&gt;
aggregate by counterparty and amount.&lt;/p&gt;

&lt;p&gt;In one case that meant nine separate transactions, each paying exactly&lt;br&gt;
486,609,148 lamports (0.486609148 SOL) to a different fresh wallet. Not&lt;br&gt;
"about 0.48". The same lamport count, nine times.&lt;/p&gt;

&lt;p&gt;Real traders sell in different sizes. Scripts round to one number.&lt;/p&gt;

&lt;h2&gt;
  
  
  The tool
&lt;/h2&gt;

&lt;p&gt;I packaged this as &lt;code&gt;solana-launch-forensics&lt;/code&gt;: a read-only CLI with no&lt;br&gt;
dependencies outside the Python standard library, no API key, and no wallet&lt;br&gt;
connection.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;scan &amp;lt;wallet&amp;gt;&lt;/code&gt; — mints created, time window, density&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;tx &amp;lt;signature&amp;gt;&lt;/code&gt; — instruction-level breakdown&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;flows &amp;lt;wallet&amp;gt; --across-tx&lt;/code&gt; — repeated equal-size payouts&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://github.com/tdkrHuang/solana-launch-forensics" rel="noopener noreferrer"&gt;https://github.com/tdkrHuang/solana-launch-forensics&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;It is MIT licensed. It reports what the chain records and does not label intent;&lt;br&gt;
repeated payouts and batched creations are patterns, not proof of wrongdoing.&lt;/p&gt;

&lt;h2&gt;
  
  
  The takeaway
&lt;/h2&gt;

&lt;p&gt;If you are reading Solana transaction logs, the number that goes wrong most&lt;br&gt;
often is the one you get by counting instructions. Attribute them first, then&lt;br&gt;
count. The two operations produce very different numbers, and only one of them&lt;br&gt;
survives review.&lt;/p&gt;

</description>
      <category>solana</category>
      <category>python</category>
      <category>security</category>
      <category>blockchain</category>
    </item>
  </channel>
</rss>
