MinishLab/semble

This page introduces an external open-source repository. It is not an HDATF product.

What it does

Semble is a code search library for agents that returns only the relevant code snippets for a natural-language query. It runs on CPU without API keys or external services, and works as an MCP server, CLI or sub-agent.

How it helps ATF

Worth comparing with tgrep for Harness executors that search code. Returning only relevant snippets instead of whole files could be a reference for keeping an agent's context small.

License

MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.

More in this category

• lfnovo/open-notebook
  Open Notebook is an open-source, self-hosted alternative to Google Notebook LM. It gathers PDFs, videos, audio and web pages, searches them by full text and vectors, lets users chat with that content and generates multi-speaker podcasts.
• zilliztech/claude-context
  An MCP plugin that adds semantic code search to Claude Code and other AI coding agents. It stores a codebase in a vector database and brings only the related code into the agent's context.
• RyanCodrai/turbovec
  A vector index written in Rust with Python bindings, built on Google Research's TurboQuant quantizer. It adds vectors without a training step, saves changes incrementally, filters by id at search time and runs fully locally.
• Graphify-Labs/graphify
  Graphify is a skill for coding assistants that maps a project's code, docs, SQL schemas, configs and PDFs into a queryable knowledge graph. Code is parsed locally with tree-sitter, and each edge is marked as extracted or inferred.
• GoogleCloudPlatform/knowledge-catalog
  Tools, agents and samples for Google Cloud Knowledge Catalog, formerly Dataplex, a data catalog and metadata platform that provides a knowledge graph over structured and unstructured data. The samples show context management, enrichment and retrieval for AI agents.

Only repositories in the ranked Trendshift lists are included, and the lists are used only to find candidates. We do not copy their ranks. Descriptions, licenses and star counts come from each GitHub repository. The notes are our own reading. We have not tested these projects, and a place on a trending list does not prove quality.