defendend/Claude-ast-index-search

Cli allows AST search, ~40-50% economy tokens. Claude/Codex/Cursor searches (suitable for any AI agent that can use Bash)

What it solves

ast-index is a structural code navigation CLI designed for large, multi-language repositories. It replaces imprecise "grep-style" text searches with an AST-aware (Abstract Syntax Tree) index, allowing developers and AI agents to navigate code by exact structure—such as classes, functions, and inheritance—rather than simple text matches.

How it works

The tool builds a local SQLite index of symbols, references, imports, modules, and dependencies. It supports a vast array of languages (including Kotlin, Java, Rust, Python, Go, and C++). Users can build the index once with rebuild and keep it current with update.

For AI agents, it provides compact, parseable structural slices of code instead of raw file dumps, which can reduce token usage by 40-50% in large repositories. It also integrates with AI tools like Claude Code, Cursor, and Gemini CLI via plugins or skills.

Who it’s for

  • Developers working in large monorepos or polyglot projects who need fast, precise navigation of symbols and dependencies.
  • AI Agent creators who want to provide their coding agents with high-precision, low-token context about a codebase.

Highlights

  • AST-Aware Navigation: Jump directly to definitions, usages, callers, and implementations across dozens of languages.
  • Agent Optimization: Returns structural slices of code to save tokens and provide better context to LLMs.
  • Git-Integrated Insights: Includes a hotspots command to rank files by churn, bugfixes, and author count to identify unstable code.
  • Symbol Graph: Builds a symbol-to-symbol dependency graph to analyze impact, find cycles, and determine the most central symbols.
  • High Performance: Significantly faster than grep for finding imports, dependents, and classes in large projects.
  • Monorepo Support: Includes specific workflows for nested project markers and cross-root workspaces.

Related

  • Project
    cocoindex-io/cocoindex-code

    An AST-based semantic code search tool that enables developers and AI agents to find code by meaning, reducing LLM token usage by up to 70%.

  • Dispatch
    SimpleEnglish Agent Skill Enables LLMs to Write in ASD‑STE100 Simplified Technical English

    The SimpleEnglish agent skill forces LLMs to produce documentation that complies with the ASD‑STE100 standard, cutting STE violations by 72.9% and shortening output across multiple Claude models.

  • Project
    NeoLabHQ/context-engineering-kit

    A plug‑in marketplace for LLM‑based coding assistants (Claude Code, Gemini, Antigravity, etc.) that supplies token‑efficient, context‑engineering patterns—reflection loops, sub‑agent judges, specification‑first development, code‑review agents, and Git helpers—to make AI‑generated code more reliable and easier to integrate into CI pipelines.

  • Project
    msitarzewski/agency-agents

    The repository provides a library of markdown‑based AI‑agent prompts (e.g., Front‑end Developer, AI Engineer, Prompt Engineer) and scripts to install them into various LLM‑powered coding assistants such as Claude Code, Cursor, Gemini CLI, etc. A companion desktop app automates installation. It’s a prompt‑template collection, not a model‑training framework.

  • Project
    Owloops/claude-powerline

    Claude Powerline is a Node‑based plugin that adds a configurable, Vim‑style status bar to Claude Code. It shows live AI usage (tokens, cost, daily/monthly budgets), Git info, model and session details, context‑window usage, rate‑limit windows, and more. Users install it via the Claude Code plugin wizard or an `npx` command, then customize segments, themes, and styles through JSON config files or the online Powerline Studio visual configurator.