I ran ripgrep, Serena, claude-context and my own tool on the same repo

说明与用途

在 Flask 3.1.3 仓库上比较 ripgrep、Serena、claude‑context 和 Semlith：ripgrep 在精确匹配时最快，Serena 擅长符号级编辑，claude‑context 需要长时间设置且可能索引不完整，Semlith 在速度和语义相关性之间取得平衡。

问题

Compare how different code‑search tools perform on the same repository when answering the same query.

结果

ripgrep was fastest for exact identifier queries; Serena excelled at symbol‑level edits; claude‑context required the longest setup and had incomplete indexing; Semlith balanced speed and semantic relevance, returning relevant code and documentation.

工作流

• Clone Flask 3.1.3 repo
• Run ripgrep search with and without identifier
• Set up Serena environment and index repository
• Query Serena with identifier and regex
• Set up claude‑context with local Milvus and Ollama
• Run claude‑context indexing and search
• Run Semlith index and search
• Compare results for specific queries
• Measure setup times and search accuracy

证据

ripgrep 15.2.0, Serena 2.1.0, claude-context 0.1.15 (with Milvus 3.0.3 and Ollama running nomic-embed-text) and semlith 0.50.0.

来源

Setup on Flask 3.1.3 on an M1 MacBook Air with 8 GB.

来源

Serena gives the agent symbol tools backed by real language servers, over 40 languages.

来源

claude-context is Zilliz's MCP server. It chunks code, embeds it and stores the vectors in Milvus, with hybrid search on top.

来源

Semlith index .
indexed 211 files (2778 chunks, 5 images) in 84.3s

来源