Skip to main content

Core Technologies

Zencoder is built on two foundational technologies:

Repo Info Agent

The Repo Info Agent indexes your entire repository to build a structural map of your codebase—architecture, patterns, dependencies, and conventions. This index powers context-aware suggestions across all agents.

Repo Info Agent

How Repo Info Agent indexes and maps your codebase

Agentic Pipeline

The Agentic Pipeline breaks coding tasks into smaller steps, routes them to specialized AI components, and validates results. It coordinates planning, code generation, testing, and error correction across a single task.

Agentic Pipeline

How the pipeline orchestrates multi-step coding tasks

Agents

Zencoder provides several specialized agents, each scoped to a specific task type:

Ask Agent

The Ask Agent answers questions about your codebase using Repo Info Agent context. It can explain functions, debug errors, and suggest approaches, but it does not modify files.

Coding Agent

The Coding Agent creates and modifies files across your codebase. It can generate multi-file implementations, search the web, run shell commands, and use MCP tools. Select it from the agent selector when you need to make changes.

Repo Info Agent

The Repo Info Agent generates a structured snapshot of your project (.zencoder/rules/repo.md) covering architecture, dependencies, and conventions. This snapshot is automatically included in every other agent’s context.

Zentester

Zentester provides two testing agents:
  • Unit Testing Agent — Generates test suites for individual components and functions using your project’s testing framework
  • E2E Testing Agent — Creates end-to-end browser tests (Playwright by default) that validate user flows

Custom AI Agents

Custom AI Agents let you define reusable agents with specific instructions, tools, and commands for recurring tasks like code review, documentation updates, or domain-specific refactoring. They can be shared across your team.

Additional Capabilities

Beyond agents, Zencoder includes:

IDE and Tool Integrations

IDE Integration

Zencoder installs as an extension from the VS Code or JetBrains marketplace. All features are accessible inside the IDE without switching to external tools.

IDE Integration

Installation steps for VS Code, JetBrains, and Android Studio

Tool Integrations

Zencoder connects to external tools through three paths:
  1. Native Integrations — Direct integration with Jira for pulling ticket context into the IDE
  2. Chrome Extension — Sends context from 20+ browser-based tools (GitHub, Sentry, Linear, etc.) to your IDE
  3. Model Context Protocol (MCP) — Open standard for connecting agents to databases, APIs, and custom tools

Tool Integrations

Setup instructions for native integrations, Chrome extension, and MCP

How Everything Fits Together

  1. Core Technologies (Repo Info Agent and Agentic Pipeline) run behind the scenes
  2. IDE & Tool Integration surfaces these capabilities inside your editor
  3. Agents provide task-specific interfaces (Ask, Coding, Testing, etc.)
  4. Capabilities are the individual operations agents can perform (generation, completion, refactoring, etc.)
Start with the Ask Agent for questions and explanations. Switch to the Coding Agent when you need to modify files.