<p>Now Claude Code remembers project context between sessions - debugging patterns, coding preferences, architectural decisions - and pulls this in the next session automatically.</p>
<p><strong>How it works:</strong></p>
<ul>
<li>Claude writes notes to the directory ~/.claude/projects/<project>/memory/</li>
<li>The main file MEMORY.md (the first 200 lines) is loaded in each session</li>
<li>Detailed notes are stored in separate files (debugging.md, api-conventions.md) and read on demand</li>
<li>You can say "remember that we use pnpm" - and it will save it</li>
</ul>
<p><strong>What it remembers:</strong></p>
<ul>
<li>Build, test, linter commands</li>
<li>Code style and naming conventions</li>
<li>Solutions to complex bugs</li>
<li>Key files and relationships between modules</li>
</ul>
<p>In addition to auto-memory, there is a hierarchy of CLAUDE.md files - from organizational level to personal settings on the project. You can even create modular rules in .claude/rules/ tied to specific paths via glob patterns.</p>
<p>Documentation: <a href="https://code.claude.com/docs/en/memory">code.claude.com/docs/en/memory</a></p>;
<p>#anthropic #claude_code #ai</p>