zeph-index

AST-based code indexing and semantic retrieval for Zeph. Always-on — no feature flag required.

Overview

Parses source files with tree-sitter to extract structured symbols (name, kind, visibility, line) via ts-query grammars, chunks them for embedding, and stores vectors in Qdrant for semantic code search. Supports Rust, Python, JavaScript, TypeScript, and Go. Generates concise repo maps that are injected into the agent context unconditionally across all LLM providers.

Key Modules

indexer — orchestrates file discovery, parsing, and embedding pipeline
retriever — semantic search over indexed symbols and chunks
store — persistence layer; vector operations go through the VectorStore trait from zeph-memory (backed by Qdrant)
repo_map — generates tree-style repository summaries using tree-sitter ts-query symbol extraction; injected into all LLM providers regardless of Qdrant availability
lsp — hover pre-filter using tree-sitter for multi-language symbol identification (Rust, Python, JS, TS, Go)
watcher — filesystem watcher for incremental re-indexing
error — IndexError error types

Supported languages

Language	Symbol extraction	Hover pre-filter
Rust	functions, structs, enums, traits, impls	yes
Python	functions, classes, methods	yes
JavaScript	functions, classes, arrow functions	yes
TypeScript	functions, classes, interfaces, types	yes
Go	functions, structs, interfaces	yes

Installation

cargo add zeph-index

[!NOTE] zeph-index does not depend on qdrant-client directly. Vector storage is delegated to zeph-memory, which owns the Qdrant client lifecycle. Repo map generation works without Qdrant — it is injected into the agent context for all LLM providers unconditionally.

Documentation

Full documentation: https://bug-ops.github.io/zeph/

License

MIT