R0MADEV

R0MADEV / lexis

Public

Lexical + structural code search MCP server for Claude Code, Cursor, Windsurf and any MCP client. Gives AI assistants 28 precise search tools so they find the right code fast 80% fewer tokens per task. No vectors, no embeddings, just ripgrep.

10
1
100% credibility
Found May 08, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

Lexis provides AI coding assistants with efficient tools to search, navigate, and understand large codebases without loading entire files.

How It Works

1
🔍 Discover Lexis

While helping your AI coding buddy explore a huge project, you hear about Lexis – a helper that makes searching code super fast and smart.

2
📦 Add it simply

With one easy step, you bring Lexis onto your computer – it handles everything by itself.

3
🔗 Link to your AI buddy

Pick your favorite AI coding tool, and Lexis connects automatically so they work together seamlessly.

4
🚀 Unlock code magic

Now your AI zips through the codebase: finds functions, traces paths, lists key spots – all without reading every line.

5
💬 Ask tough questions

Chat with your AI about bugs or features, like 'fix the config reload issue' – get spot-on answers in seconds.

6
📝 Save your insights

Your AI jots down key discoveries as notes, ready for your next session weeks later.

🎉 Conquer any codebase

You and your AI tackle massive projects effortlessly, saving time and getting things done right.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is lexis?

Lexis is a TypeScript MCP server that equips AI coding assistants like Claude Code, Cursor, and Windsurf with 28 lexical and structural search tools for ripping through large codebases. Using ripgrep plus lightweight AST parsing—no vectors or embeddings—it delivers precise symbol lookups, call chains, and file slices, slashing token use by 80% on tasks like bug hunts. Install globally via npm, and it auto-indexes projects, persists branch-specific notes, and prioritizes src/ over tests/.

Why is it gaining traction?

It stands out with lexical and structural accuracy that beats generic file reads or embedding-based tools, offering modes like "compact" snippets or "count" for token thrift without losing precision—perfect for lexical vs structural ambiguity in multi-language stacks. Developers hook on the zero-setup global install, CLI for manual queries (`lexis ask "bug in reloadCache"`), and tools like `call_chain` or `impact_analysis` that reveal flows instantly. No external services means it runs offline, bundling ripgrep for all platforms.

Who should use this?

Backend engineers debugging 100k+ LOC PHP, Go, or Elixir repos, especially telecom stacks with Kamailio or Asterisk configs. AI-heavy workflows in Cursor/Claude Code where token budgets kill productivity on feature branches. Skip if your repo fits in one prompt or you need full LSP type-checking.

Verdict

Early gem at 10 stars and 1.0% credibility—untested beyond Claude Code, but 68 tests, strict TS, and real-world validation on 500k-line codebases make it worth a spin for MCP users. Try the global install if you're token-constrained; contribute parsers for your DSL.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.