How does hybrid search work?

Hybrid search uses Reciprocal Rank Fusion to combine the precision of keyword matching (BM25) with the conceptual understanding of semantic vector search.

Can I filter results by project?

Yes, you can use the --workspace flag to limit your search to specific project paths or directories to find project-specific decisions.

How do I see more context from a search result?

You can use the 'cass expand' or 'cass view' commands to see the lines surrounding a specific match or the full source of the historical session.

What agents does Session Search support?

It supports sessions from Claude Code, Cursor, Codex, Gemini CLI, and other major coding assistants, indexing over 2,600 conversations.

Is this skill token-efficient?

Yes, it includes features like --max-tokens, --fields minimal, and truncation options to ensure search results don't overwhelm Claude's context window.

Coding Agent Session Search

Name: Coding Agent Session Search
Author: ASRagab

byASRagab

0•

生产力与工作流

Searches and retrieves historical coding patterns, solutions, and decisions across thousands of previous AI agent sessions.

Session Search (CASS) enables Claude to index and query past conversations from multiple AI agents including Claude Code, Cursor, and Gemini CLI. It provides a powerful retrieval system using lexical, semantic, and hybrid search modes, allowing developers to recall specific fixes, architectural decisions, or code snippets from their collective history. By bridging the gap between current tasks and past successes, it prevents redundant work and facilitates better knowledge management within the development workflow.

主要功能

01Hybrid search combining BM25 keyword matching and vector-based semantic retrieval.

02Context expansion to view surrounding messages and source files from past sessions.

03Multi-agent support for Claude Code, Cursor, Codex, and Gemini CLI histories.

04Advanced filtering by workspace, agent type, and specific metadata.

05Token-efficient output controls to manage context window usage effectively.

060 GitHub stars

使用场景

01Finding an exact error fix implemented in a previous project session.

02Recalling the design rationale for a specific API architecture discussed weeks ago.

03Aggregating and storing reusable code patterns across multiple different AI coding agents.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add asragab/asragab-claude-marketplace session-search

For use in Claude.ai and ChatGPT

主要功能

01Hybrid search combining BM25 keyword matching and vector-based semantic retrieval.

02Context expansion to view surrounding messages and source files from past sessions.

03Multi-agent support for Claude Code, Cursor, Codex, and Gemini CLI histories.

04Advanced filtering by workspace, agent type, and specific metadata.

05Token-efficient output controls to manage context window usage effectively.

060 GitHub stars

使用场景

01Finding an exact error fix implemented in a previous project session.

02Recalling the design rationale for a specific API architecture discussed weeks ago.

03Aggregating and storing reusable code patterns across multiple different AI coding agents.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add asragab/asragab-claude-marketplace session-search

For use in Claude.ai and ChatGPT