Multi-Repository Search and Zen Rules Platform Update

In July we released two major features: Multi-Repository Search and Zen Rules. These features enable better contextual understanding across complex enterprise codebases by allowing agents to search across multiple repositories and apply project-specific instructions. Multi-repo search addresses the challenge of understanding dependencies and patterns in distributed systems, while Zen Rules provides a mechanism for encoding architectural decisions and coding standards directly into the AI’s context. We also introduced enhanced agents (Repo-Info, Q&A, E2E Testing), universal image attachment support, and numerous platform improvements for better stability and performance.

New Features

We’re excited to introduce Multi-Repository Search, enabling AI agents to understand and work across your entire organization’s codebase:
  • Index multiple repositories through the web admin panel at auth.zencoder.ai
  • Cross-repository code understanding allows agents to find patterns, implementations, and dependencies across all your indexed repositories
  • Automatic context detection means agents intelligently search relevant repositories when you reference other services or projects

Learn More About Multi-Repository Search

Set up and configure multi-repository indexing to enable AI-powered search across your entire codebase

Zen Rules - Project-Specific AI Instructions

Introducing Zen Rules, our powerful new system for creating custom context and instructions that enhance how AI agents understand your project:
  • Project-level rules stored in .zencoder/rules/*.md files that are committed to your repository for team sharing
  • Conditional application based on file patterns using glob expressions, or always-on rules for project-wide standards
  • Three-tier hierarchy combining Zen Rules, personal AI instructions, and repo.md for comprehensive context management
  • @mention support allows you to manually reference specific rules in your chat for targeted guidance

Explore Zen Rules

Learn how to create custom rules that enforce your team’s coding standards and architectural patterns

Enhanced Repo-Info Agent

The Repo-Info Agent has been significantly improved to provide deeper insights into your codebase:
  • Better project analysis with more comprehensive understanding of build systems, dependencies, and architectural patterns
  • Optimized context generation reduces the number of LLM calls needed for subsequent agent requests
  • Automatic migration moves repo.md from the old /docs location to the new .zencoder/rules/ directory
  • Foundation for all agents as the generated context is automatically included in every agent request

Learn About Repo-Info Agent

Discover how the Repo-Info Agent analyzes your project structure and generates comprehensive context for all other agents

Q&A Agent

We’ve replaced the legacy basic chat with our new Q&A Agent, optimized for quick questions and code explanations:
  • Lightweight interface designed for rapid responses to coding questions
  • Codebase-aware leveraging Repo Grokking to understand your specific project context
  • Streamlined for simplicity with limited tool access for faster, focused interactions
  • Easy agent switching through the new agent selector UI

Explore Q&A Agent

Learn about our streamlined Q&A Agent designed for quick coding questions and explanations

Image Attachments for All Users

Visual context is now universally available across both VS Code and JetBrains IDEs:
  • Universal access means that image attachment functionality is no longer limited and is available for all users across all supported platforms
  • Design-to-code workflow has been streamlined, allowing you to attach UI mockups or screenshots and ask agents to build matching components with high accuracy
  • Repository images stored in your project can be referenced for visual context, enabling agents to understand and implement visual designs directly from your codebase assets

E2E Testing Agent Enhancements

The E2E Testing Agent has evolved to become framework-agnostic:
  • Multi-framework support now enables you to generate tests for Cypress, Playwright, or Selenium based on your project’s requirements and preferences
  • Flexible test generation approach means you simply specify your preferred framework in the prompt, and the agent adapts its output accordingly
  • Maintained intelligence ensures that all the powerful test creation, exploration, and fixing capabilities that made our E2E agent valuable now work seamlessly across all supported frameworks

Learn About E2E Testing Agent

Discover how to generate end-to-end tests using Cypress, Playwright, or Selenium with our intelligent testing agent

Improvements and Bug Fixes

Agent Experience Enhancements

  • New agent selector UI provides quick access to different agents with a redesigned dropdown interface
  • Background shell execution allows agents to run long-running commands without blocking the interface
  • File Diagnostic tool is now available for custom agents, expanding their capabilities
  • Better file mention suggestions with improved logic based on user feedback for more intuitive file referencing
  • Expandable code blocks in chat for better readability with large code snippets
  • Agent continuation shows an improved message when agents reach max steps with option to continue the request

User Interface Improvements

  • Redesigned VS Code status bar with more visible Zencoder icon and better integration
  • Updated sidebar menu in the admin UI with improved organization and new profile dropdown for easier account management
  • Premium LLM usage display clearly shows call consumption for users on new pricing plans
  • Better code diff copying now only copies added lines when copying from diffs, not the entire block
  • Enhanced chat input UI with cleaner design and better visual hierarchy

Platform Stability and Performance

  • SSE/HTTP streaming support for remote MCP servers without OAuth (VS Code, coming to other platforms)
  • Requirements tool functionality helps agents better understand user needs before proceeding - now with improved UI
  • Less frequent update notifications reduces interruption while keeping you informed of important updates
  • repo.md migration automatically moves from deprecated /docs to new /rules directory structure

Bash/PowerShell/WSL Improvements

  • PowerShell support with PS1 files now properly executing shell commands on Windows
  • WSL compatibility adds support for WSL and other interpreter types on Windows
  • Bash tool stability fixed job cancellation issues when Kotlin plugin is not installed in JetBrains

Bug Fixes

  • Shell tool freezing resolved cases where shell tool would freeze in certain scenarios
  • Shell pagination now properly avoids pagination in command output for cleaner results
  • Code generation window fixed size adjustment issues when pasting text in JetBrains
  • Q&A Agent tools fixed issue where Q&A Agent was using tools it shouldn’t have access to
  • File attachment labels updated to better reflect actual capabilities and limitations
  • LLM call counter now properly displays consumption after clicking Stop
  • Various UI glitches resolved multiple interface inconsistencies across platforms

Documentation Updates

We’ve significantly expanded and improved our documentation to support the new features and provide better guidance:

New Documentation Pages

  • Agents Overview serves as the new landing page for all agents, providing a comprehensive guide to accessing and understanding each agent’s capabilities
  • Q&A Agent documentation introduces our new lightweight chat interface, explaining how it differs from the Coding Agent and when to use each
  • Multi-Repository Search guide provides detailed setup instructions, best practices, and examples for indexing and searching across multiple repositories
  • Zen Rules documentation explains the three-tier hierarchy of AI instructions, with practical examples and guidelines for creating effective project-specific rules
  • Refund Policy FAQ offers clear information about subscription management, cancellations, and our transparent refund policies

Enhanced Existing Documentation

New Tutorials and Visual Updates

  • Generate Codebase Architecture Diagrams tutorial shows how to create Mermaid diagrams to visualize repository structure and dependencies
  • Image prompts documentation updated to explain the new universal image attachment capabilities
  • Agent selector images refreshed across all agent documentation to show the new dropdown UI and improved user experience
  • Multi-repo search guidelines enhanced with specific examples and best practices for effective cross-repository searches

Version History

  • 2.17.0 (July 23, 2025)
  • 2.16.0 (July 15, 2025)