The latest articles on DEV Community by MD (@md_565627). https://dev.to/md_565627 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3960377%2Ff90d7d13-3f38-4d11-9f55-d0738079ec31.png https://dev.to/md_565627 en MD Sat, 30 May 2026 19:01:10 +0000 https://dev.to/md_565627/how-i-cut-ai-coding-token-costs-by-96-with-ast-compression-27hn https://dev.to/md_565627/how-i-cut-ai-coding-token-costs-by-96-with-ast-compression-27hn <p>If you use Claude Code, Cursor, or Aider - you feel the pain.</p> <p>You ask to fix a bug and it dumps 15 files into context. 45K tokens wasted.</p> <p>I built <strong>PMC Engine</strong> - AST-level compression that scores every symbol and sends only what's needed.</p> <h2> The Numbers </h2> <p>FastAPI (48 files, 33K LOC, DeepSeek V4 Flash):</p> <ul> <li>Simple fix: 45K -&gt; 7.1K (84%)</li> <li>Refactor: 85K -&gt; 7.8K (91%)</li> <li>Complex: 148K -&gt; 7.7K (95%)</li> <li><strong>Avg: 91.8% reduction, 100% quality, &lt;5ms</strong></li> </ul> <h2> Quick Start </h2> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>pip <span class="nb">install </span>pmc-engine pmc index ./my-project pmc serve <span class="nt">--port</span> 8080 </code></pre> </div> <p>MIT: <a href="https://github.com/mdayan8/pmc-engine" rel="noopener noreferrer">https://github.com/mdayan8/pmc-engine</a></p> claude opensource python productivity