archive-search-claudeaceae-samara-main-2

npx skills add majiayu000/claude-skill-registry --skill archive-search-claudeaceae-samara-main-2 --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

# Archive Search Search through raw Claude Code session transcripts to recover technical details, reasoning traces, and thinking blocks that were compressed during episode distillation. ## What You'll Get **The archive contains:** - ✅ Complete thinking blocks (raw reasoning process) - ✅ User messages (original intent and questions) - ✅ Assistant explanations (first 1000 chars) - ❌ Tool results (filtered out - too noisy) - ❌ System messages (no semantic value) **Best for:** - Technical archaeology: "How did we implement X?" - Decision rationale: "Why did we choose Y over Z?" - Reasoning evolution: "How has our approach to X changed?" - Debugging context: "What was the exact error we saw?" ## How Archive Differs from Regular Memory | Memory Type | Content | Use Case | |-------------|---------|----------| | **Episodes** (`/episode`) | Curated summaries (2-8 bullets per section) | What happened today | | **Learnings** (`/memory`) | Distilled insights | General knowledge | | **Archive** (`/archive-search`) | Raw transcripts with thinking blocks | Technical depth, exact discussions | The archive captures ~5-10% of raw transcript bytes by selecting only high-value content (thinking blocks

Build initial index (takes 5-10 minutes for 90 days)
Check stats
Test search