skills.rest
obra
obraCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

Remembering Conversations

Search archived Claude Code conversations via semantic similarity or exact text matching.

41|27|Updated Oct 6, 2025
Search Tags:#context#knowledge management#AI assistant#semantic search#search#memory#conversation history
One-click install
npx skills add https://github.com/obra/clank --skill remembering-conversations
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Remembering Conversations
Source: https://github.com/obra/clank/tree/main/skills/collaboration/remembering-conversations
Command: npx skills add https://github.com/obra/clank --skill remembering-conversations

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @anthropic-ai/claude-agent-sdk, @xenova/transformers, better-sqlite3, sqlite-vec, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Losing track of past discussions, decisions, or code patterns in previous AI conversations can lead to duplicated effort and inconsistent solutions. This skill provides a powerful search mechanism to instantly retrieve relevant context from your entire conversation history, ensuring you always build upon past knowledge.

Core Features & Use Cases

  • Semantic & Text Search: Find conversations using natural language queries or exact string matches (e.g., for Git SHAs).
  • Automatic Indexing: Installs a hook to automatically index new conversations after each session, keeping your knowledge base up-to-date.
  • AI Summaries & Snippets: Provides AI-generated summaries and relevant snippets for quick context, saving you from reading full conversations.
  • Use Case: When your human partner asks about "how we handled auth errors in React Router last month," you'll use this skill to semantically search your past conversations, quickly retrieving the exact discussion and solution.

Quick Start

Install auto-indexing hook (one-time setup)

~/.claude/skills/collaboration/remembering-conversations/tool/install-hook

Index existing conversations

~/.claude/skills/collaboration/remembering-conversations/tool/index-conversations --cleanup

Search for a topic

~/.claude/skills/collaboration/remembering-conversations/tool/search-conversations "React Router authentication errors"

Frequently Asked Questions about Remembering Conversations

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search across my past Claude Code conversations?▼

Semantic search across archived conversations uses natural language queries to find relevant prior discussions. Install the auto-indexing hook, index your existing conversations, then search using plain English phrases like "React Router authentication errors" to retrieve matching sessions with AI-generated summaries and snippets.

Can I find specific code or decisions from months ago in my conversation history?▼

Yes. The skill indexes conversations with both semantic similarity and exact text matching, letting you locate specific Git SHAs, code patterns, or past decisions. A local vector index with 384-dimensional embeddings powers instant retrieval across multiple conversations and time ranges.

How does automatic conversation indexing work?▼

A one-time hook installation automatically indexes new conversations after each session, maintaining an up-to-date knowledge base without manual effort. The local sqlite-vec vector index and embedding model run concurrency-controlled indexing, with permanent archiving that includes per-conversation summaries, project context, and exchange details.

What's the best way to organize and retrieve decisions from multiple AI conversations?▼

Semantic search with AI-generated summaries lets you organize and instantly recall decisions across conversations by natural language intent, not folder structure. Query your archive with phrases describing what you're looking for—the skill returns relevant context, snippets, and metadata without reading full transcripts.

Do I need to set up a database to search my conversation history?▼

No manual database setup required. The skill uses a local sqlite-vec vector index with a 384-dimensional embedding model installed automatically. Permanent archiving with summaries happens in place; CLI tools handle indexing and searching without additional configuration beyond the one-time hook installation.

Can exact text matching find conversations if semantic search doesn't work?▼

Yes. The skill supports both semantic similarity and exact string matching, so you can locate conversations by Git SHAs, specific code snippets, or exact phrases when natural language queries don't yield results.

Related Skills

Explore More Data & Analytics Skills→
christophevgchristophevg

transcribe-session

Transcribe conversations into structured transcripts with YAML frontmatter and synthesis sections.

Community
Intermediate
devopstalesdevopstales
1

documentation-and-adrs

Generate README files, inline code documentation, and architectural decision records.

Community
Advanced
zangjeicyzangjeicy

chroma

Manage vector embeddings in an open-source database for RAG and semantic search.

Community
Intermediate
openakitaopenakita
2.0k

search-memory

Search and retrieve memories by keywords and optional type filters.

Community
Basic