Dicklesworthstone/coding_agent_session_search

Unified TUI and CLI to index and search your local coding agent session history across 11+ providers (Codex, Claude, Gemini, Cursor, Aider, etc.)

What it solves

Coding agents (like Cursor, Claude Code, and Aider) generate vast amounts of conversation history, but this data is typically fragmented across different proprietary formats and storage locations. This makes it difficult to find specific solutions or context from past sessions across different tools. cass provides a unified, high-performance interface to index and search this local history.

How it works

cass normalizes session data from various agents into a common schema and stores it in a local SQLite database. It uses a multi-tiered search approach:

  • Lexical Search: Uses BM25 full-text search with edge N-grams for instant, prefix-based matching of exact terms and code symbols.
  • Semantic Search: An optional, opt-in feature that uses a local MiniLM model (via frankensearch) to perform vector similarity searches for conceptual queries without needing a network connection.
  • Hybrid Search: The default mode, which combines lexical and semantic results using Reciprocal Rank Fusion (RRF) to balance precision and recall.

It features a Terminal User Interface (TUI) for humans and a "Robot Mode" (CLI with JSON output) specifically designed for other AI agents to query the history.

Who it’s for

  • Individual developers who want a single searchable archive of their AI-assisted coding sessions.
  • Teams looking to share institutional knowledge across different tool preferences.
  • AI agents that need to access a user's historical context to improve current problem-solving.

Highlights

  • Broad Compatibility: Aggregates history from Codex, Claude Code, Gemini CLI, Cursor, Aider, GitHub Copilot, and many others.
  • Sub-60ms Latency: High-speed "search-as-you-type" experience powered by a Rust-native backend.
  • Privacy-First: All indexing and inference happen locally; no data is sent to external servers.
  • Robot Mode: A dedicated API surface with stable JSON contracts for seamless integration with other AI agents.
  • Atomic Updates: Ensures index integrity using atomic swaps, preventing corrupted search results during background indexing.