Global configuration model with validation.

Stateless core indexing service orchestrating all components.

Immutable entity representing a code component.

Qdrant vector database implementation.

Result of a storage operation.

OpenAI embeddings with retry logic and rate limiting.

Result of an embedding operation.

Result of parsing a code file.

Represents a chunk of entity content for vector storage in progressive disclosure architecture.

Parser for Python files using Tree-sitter and Jedi.

Unified configuration loader with project-level support.

Represents a point in vector space.

Parser for Markdown documentation files with intelligent chunking.

Immutable relationship between two entities.

Unified watcher class using IndexingEventHandler for reliable file watching.

Registry for managing multiple code parsers.

Manages project-specific configuration.

Parser for Claude Code JSONL conversation files.

Manages Memory Guard bypass state with simple on/off commands.

Background service for continuous file watching across multiple projects.

Single message in a Claude Code conversation.

Summarizes Claude Code conversations using OpenAI API.

Fast, deterministic embedder for testing.

Manage git hooks for automatic indexing on commits.

File system event handler with debouncing and batch processing.

Result of an indexing operation.

Result of chat conversation summarization.

Vector store with result caching.

Generates HTML reports combining GPT analysis with conversation display.

Represents a point with both dense and sparse vectors for hybrid search.

Extract semantic observations from code elements.

Complete project configuration.

Voyage AI embeddings with retry logic and rate limiting.

Parse CSS/SCSS with tree-sitter.

Parse CSV files for basic structure.

Wrapper that adds caching to any embedder.

Represents chat data as a chunk for v2.4 pure architecture.

Complete conversation from Claude Code.

Metadata extracted from a chat conversation.

Centralized code analysis with optimized compiled patterns.

Mechanical summary of changes (Meta's approach)

Registry for creating and managing embedders.

Enhanced orphan cleanup for hash-based storage scenarios

Extract entities from code content.

Processor for entity metadata chunks.

Simple file change coalescer with background timer for automatic processing.

File patterns for include/exclude.

Context information from Git+Meta content analysis.

Parse HTML with tree-sitter for structure and components.

Parse INI/Config files.

Processor for implementation chunks.

Indexing behavior configuration.

Parse JSON with tree-sitter for structural relation extraction.

Parse JS/TS files with tree-sitter, optional TSServer for semantics.

JavaScript parser specific configuration.

Markdown parser configuration.

Comprehensive code quality gate - checks duplication, logic, flow integrity, and feature preservation.

Parser-specific configuration.

Context information for content processing operations.

Result of content processing operations.

Project metadata.

Processor for relation chunks with smart filtering.

Processes only relations involving changed entities

Registry for creating and managing vector store instances.

Parse plain text files with configurable chunking.

Orchestrates unified content processing pipeline.

Parse YAML configuration files.

Async debouncer with coalescing for file system events.

BM25 sparse embeddings for keyword-based semantic search. Generates sparse vectors compatible with Qdrant format using the bm25s library.

Abstract base class for code parsers.

Mixin for content-addressable storage functionality

Base class for content processing with deduplication.

Represents a change layer in the immutable diff system

Manages immutable diff layers for efficient change tracking

Abstract base class for text embedding generators.

Factory for creating entities with consistent patterns.

Types of entities that can be extracted from code.

File inclusion/exclusion patterns.

Indexing behavior configuration.

JSON parser configuration.

JavaScript/TypeScript parser configuration.

Vector store with automatic collection management.

Project metadata.

Python parser specific configuration.

Collects comprehensive statistics from Qdrant vector database.

Represents a relation as a chunk for v2.4 pure architecture.

Factory for creating relations with consistent patterns.

Types of relationships between entities.

Base class for embedders that support retry logic.

Intelligent watcher completion detection via log analysis.

Test complete indexing workflows.

Custom tests for precise incremental indexing behavior verification.

Test complete CLI workflows.