# session-indexer session-indexer est un outil pour développeurs hébergé sur GitHub qui indexe les transcriptions de sessions Claude Code dans SQLite et permet de les rechercher par sens. Il est conçu pour retrouver l’historique JSONL par projet, avec FTS5 en solution de repli lorsque la recherche basée sur des embeddings n’est pas disponible. - Findings24: - Site web: - Tarifs: Freemium - Catégories: Moteur de Recherche IA, Outils Dev IA ## Qu'est-ce que session-indexer ? Session-indexer is a developer tool for searching Claude Code session history on a per-project basis. It indexes JSONL transcripts into SQLite and then lets users retrieve past sessions with semantic search powered by bge-m3 embeddings through Ollama, with FTS5 available as a fallback. The project is presented as a GitHub repository and appears to be aimed at developers who want a local, queryable record of Claude Code sessions rather than a generic note-taking or search app. Its core purpose is to make past sessions easier to find by turning transcript files into a searchable index. ## Que peut faire session-indexer ? ### Indexes JSONL transcripts into SQLite The project stores session history in SQLite, which makes the transcript data queryable in a structured local database instead of leaving it only as raw files. ### Semantic retrieval with bge-m3 embeddings The page description states that search is performed using bge-m3 embeddings via Ollama, which indicates embedding-based matching for searching by meaning rather than exact text alone. ### FTS5 fallback search When semantic retrieval is not used or not available, the project can fall back to SQLite FTS5, providing a text-search path over indexed transcript content. ### Per-project session scope The product is described as per-project search over Claude Code history, suggesting the index is organized around individual project contexts rather than one global transcript pool. ### Repository includes project docs The repository contains a docs directory and a .claude directory, indicating that usage and project-specific configuration/documentation are part of the codebase. ## Cas d'utilisation ### Find a prior Claude Code discussion in one project A developer can search session history for a specific project when they need to recover a decision, command, or troubleshooting thread from earlier Claude Code work. ### Locate information by meaning rather than exact wording Semantic search is useful when the user remembers the topic of a session but not the exact phrase used in the transcript. ### Search when embedding retrieval is not available If the embedding path is not in use, FTS5 provides a fallback for users who still need text-based retrieval over the indexed transcripts. ### Maintain a structured archive of transcripts Developers can keep JSONL transcripts in SQLite so session history is easier to inspect and reuse later. ## Questions fréquentes ### What does session-indexer index? It indexes JSONL transcripts from Claude Code session history into SQLite. ### How does search work? The project uses bge-m3 embeddings through Ollama for semantic retrieval, with FTS5 as a fallback. ### Is this a cloud service or a local tool? The source describes a repository that indexes data into SQLite and does not indicate a hosted SaaS product. ### Does it support non-semantic search? Yes. The page explicitly mentions FTS5 fallback. ### Where can I learn more about usage? The repository includes docs and project files, but the provided page text does not show a full usage guide. ## Informations rapides - **Category**: Developer Tool - **Primary task**: Semantic search over Claude Code session history - **Data format**: JSONL transcripts - **Storage**: SQLite - **Search method**: bge-m3 embeddings via Ollama, with FTS5 fallback - **Source domain**: github.com