What is the difference between semantic and text search in this skill?

Semantic search finds conversations based on meaning and conceptual similarity, while text search looks for exact character-for-character matches like error codes or Git SHAs.

How do I use this skill manually in the terminal?

You can run the 'search-conversations' command directly from your shell to query your conversation history without starting a full Claude session.

Can I search for conversations from a specific time period?

Yes, the tool supports --after and --before flags, allowing you to filter results by specific YYYY-MM-DD dates.

Does searching past conversations consume a lot of tokens?

When used with subagents as recommended, this skill can save 50-100x on context costs by only bringing the relevant search results into the main session.

Conversation History Search

Name: Conversation History Search
Author: lifangda

bylifangda

•

협업 도구

Searches archived Claude Code conversations using semantic and text matching to retrieve past decisions, patterns, and facts.

소개

Conversation History Search is a specialized capability that provides Claude with long-term memory by indexing and retrieving data from past interactions. By utilizing both vector-based semantic similarity and exact string matching, it allows users and AI subagents to recall historical context, architectural decisions, and debugging solutions without repeating previous discussions. This skill is particularly valuable for maintaining consistency across long-running projects and reducing context window usage by fetching only relevant historical snippets.

주요 기능

Dual-mode retrieval using semantic similarity and exact text matching
Advanced filtering by date ranges and result limits
Optimized subagent integration for high-efficiency context retrieval
Direct CLI tool for manual historical lookups outside of active sessions
Detailed result reporting including similarity percentages and conversation summaries
16 GitHub stars

사용 사례

Retrieving specific architectural decisions or design patterns established in previous months
Referencing past debugging steps for recurring or similar issues to accelerate resolution
Fact-checking project history when a partner mentions 'we discussed this before'

소개

주요 기능

Dual-mode retrieval using semantic similarity and exact text matching
Advanced filtering by date ranges and result limits
Optimized subagent integration for high-efficiency context retrieval
Direct CLI tool for manual historical lookups outside of active sessions
Detailed result reporting including similarity percentages and conversation summaries
16 GitHub stars

사용 사례

Retrieving specific architectural decisions or design patterns established in previous months
Referencing past debugging steps for recurring or similar issues to accelerate resolution
Fact-checking project history when a partner mentions 'we discussed this before'