---
title: "Overview"
description: "Overview of Verdent's AI-powered development capabilities"
---

Verdent for VS Code combines advanced AI capabilities with professional development workflows. This guide introduces the core features that enable AI-assisted coding, from requirement engineering through automated quality assurance.

###  What You'll Learn

- Core workflow capabilities (Professional Plan, Code Diff, Verify)
- Context awareness and specialized sub-agents
- Collaboration modes and extensibility

---

Verdent's workflow is built around three core phases:

- **Plan** - Clarify requirements with Plan Mode
- **Code** - Review and refine with Code Diff for reliable delivery
- **Verify** - Automate testing with Verify tools and catch issues early

These features work together to ensure precision and control throughout your development process.

---

## Core Workflow Features

<CardGroup cols={2}>
  <Card title="Plan-Code-Verify Workflow" icon="crosshairs" href="verdent-for-vscode/core-features/plan-code-verify-loop">
    Precise control over requirements, implementation, and validation through  verify subagent and browser tools
  </Card>
  <Card title="Professional Plan" icon="diagram-project" href="/docs/verdent-for-vscode/core-features/professional-plan">
    Transform ideas into production-ready implementation plans with intelligent requirement clarification
  </Card>
  <Card title="Code Diff" icon="code-compare" href="/docs/verdent-for-vscode/core-features/code-diff">
    Enterprise-grade code review with natural language summaries and senior engineer-level analysis
  </Card>
  <Card title="Clarification Loop" icon="message-dots" href="/docs/verdent-for-vscode/core-features/clarification-loop">
    Smarter planning through continuous discussion and refinement of requirements
  </Card>
  <Card title="Plan Rules" icon="list-check" href="/docs/verdent-for-vscode/core-features/plan-rules">
    Customize how clarifications and plans are presented based on your role and professional background
  </Card>
</CardGroup>

---

## Additional Capabilities

### Context Awareness: Deep Codebase Understanding

Verdent's context management system enables comprehensive project comprehension:

#### Massive Context Window

- **1M Token Capacity** - Ingest entire medium-sized codebases (~750,000 words or 3,000+ files)
- **Smart Context Loading** - Automatically prioritizes relevant files based on task context
- **Sub-Agent Context Optimization** - Delegates specialized tasks to focused sub-agents

#### Adaptive Learning

- **Convention Detection** - Learns project-specific patterns (naming, file organization, error handling)
- **Style Mimicry** - Generates code matching existing style (indentation, brace placement, comments)
- **Library Awareness** - Recognizes frameworks in use, preferring them over new dependencies

#### Cross-File Coherence

- **Dependency Tracking** - Understands imports, exports, and module relationships
- **Impact Prediction** - Identifies components affected by proposed changes
- **Consistency Enforcement** - Ensures modifications align with existing architecture

---

### Specialized Sub-Agents: Division of Labor

Verdent orchestrates specialized AI agents optimized for specific development tasks:

<Tabs>
  <Tab title="Explorer Agent">
    **Purpose:** Fast, evidence-based codebase navigation and analysis

    **Capabilities:**

    - **Pattern Matching** - Locate files using glob patterns (e.g., all TypeScript files, backend API files)
    - **Semantic Search** - Find code by functionality (e.g., "where is authentication middleware implemented?")
    - **Multi-Location Synthesis** - Aggregate information from multiple files

    **Thoroughness Levels:**

    - **Quick:** Basic pattern matching for fast answers
    - **Medium:** Broader search with contextual confirmation
    - **Very Thorough:** Exhaustive scan with variant checking and cross-references

    **Use Cases:** "Find all database query functions", "Locate configuration loading logic", "How does the app handle errors?"
  </Tab>
  <Tab title="Verifier Agent">
    **Purpose:** Rapid code quality checks and validation

    **Capabilities:**

    - **Lint Checks** - ESLint, Pylint, Rubocop, etc.
    - **Type Validation** - TypeScript, mypy, Flow type checking
    - **Fast Test Execution** - Targeted unit tests with under 30s budget
    - **Diff-Focused Verification** - Primarily validates changed code for efficiency

    **Fail-Fast Philosophy:** Returns structured error reports on first real issue, avoiding time waste

    **Use Cases:** Pre-commit checks, post-fix validation, quick sanity tests
  </Tab>
  <Tab title="Code Reviewer Agent">
    **Purpose:** Comprehensive code quality, security, and maintainability audits

    **Capabilities:**

    - **Security Analysis** - Vulnerability scanning, credential exposure detection, injection risk assessment
    - **Architecture Review** - Design pattern validation, SOLID principle adherence, anti-pattern detection
    - **Performance Analysis** - Algorithmic complexity, resource leaks, optimization opportunities
    - **Documentation Review (for changed files)** - flags unclear comments or obvious doc inconsistencies

    **Review Depth:** More thorough than Verifier, suitable for pre-merge PR reviews

    **Use Cases:** Feature completion review, security audit, refactoring validation
  </Tab>
</Tabs>

---

### Flexible Collaboration Modes

Choose the level of autonomy that fits your workflow:

- **Auto Run Mode** - Executes tasks autonomously while notifying you of potentially risky actions
- **Manual Accept Mode** - Requires your approval for every change before execution
- **Skip Permissions Mode** - Fully autonomous execution, including risky operations (advanced users only)

See [Execution Modes & Permissions](/docs/verdent-for-vscode/execution-modes/overview) for detailed mode documentation.

---

### MCP (Model Context Protocol) Integration

Enables interoperability with external tools and services:

- Extends functionality through existing toolchains and custom plugins
- Works seamlessly with sub-agents to support distributed task execution
- Supports integration with external APIs, databases, and development tools

See [Integration & Extensions](/docs/verdent-for-vscode/advanced-features/integrations) for MCP setup and configuration.

---

### Additional Features

<Tabs>
  <Tab title="Auto Model Selection">
    **Intelligent Model Optimization:**

    Verdent automatically selects the most suitable AI model for each task based on complexity, performance requirements, and cost considerations.

    **Features:**

    - **Task Analysis** - Evaluates task complexity to determine optimal model
    - **Performance Balancing** - Weighs speed, accuracy, and cost trade-offs
    - **Context-Aware Selection** - Adjusts model choice based on project size and requirements
    - **Cost Optimization** - Uses lighter models for simple tasks, reserves powerful models for complex operations

    **Benefits:** Maximizes efficiency while minimizing credit usage, ensuring you get the best results without overspending.
  </Tab>
  <Tab title="Context Referencing">
    **Precise Context Control:**

    Attach specific files, folders, or code sections directly in chat using `@` mentions to provide targeted context for AI assistance.

    **How It Works:**

    - Type `@` in chat to see a list of available files and folders
    - Select specific files to include in the conversation context
    - Reference entire directories for broader context
    - Mention specific code sections or documentation pages

    **Use Cases:**

    - Focus AI on specific modules when debugging
    - Include configuration files when discussing setup
    - Reference related components when implementing features
    - Provide documentation context for accurate guidance
  </Tab>
  <Tab title="Visual Support">
    **Multi-Modal Input:**

    Upload or paste images directly into chat to communicate visual requirements, UI designs, or debugging scenarios.

    **Supported Use Cases:**

    - **UI/UX Design** - Upload mockups, wireframes, or design screenshots for pixel-perfect implementation
    - **Bug Reports** - Share error screenshots or visual glitches for faster diagnosis
    - **Frontend Development** - Provide design references for accurate styling and layout
    - **Documentation** - Include diagrams, flowcharts, or architecture visualizations

    **Technical Specifications:**

    - **Supported Formats:** PNG, JPG, JPEG, GIF, BMP, WebP
    - **Maximum Resolution:** 2000×2000 pixels
    - **File Size Limit:** 5MB (automatically compressed if exceeded; error if still \>5MB after compression)

    **Benefits:** Eliminates ambiguity in visual requirements and accelerates frontend development workflows.
  </Tab>
  <Tab title="Project History">
    **Session Management:**

    Access and manage your conversation history to review past interactions, decisions, and implementation details.

    **Features:**

    - **Session Logs** - Complete record of all conversations and AI responses
    - **Repository Storage** - Logs saved alongside your project for easy access and version control
    - **Session Clearing** - Clear history when starting fresh or switching contexts
    - **Decision Trail** - Review reasoning behind past implementation choices

    **Benefits:** Maintain continuity across sessions, audit AI recommendations, and track project evolution over time.
  </Tab>
  <Tab title="User Center">
    **Account Management:**

    Centralized hub for managing your Verdent account, credits, and plugin settings.

    **Available Controls:**

    - **Credit Tracking** - Monitor daily credit usage and remaining balance
    - **Account Information** - View current subscription plan and profile details
    - **Sign-Out Management** - Securely log out from the plugin or switch accounts
    - **Usage Monitoring** - Track credit consumption patterns

    **Access:** Available through the Verdent sidebar in VS Code.
  </Tab>
  <Tab title="Feedback">
    **Product Improvement:**

    Submit feedback, bug reports, and feature requests directly from VS Code to help improve Verdent.

    **Feedback Options:**

    - **Text Notes** - Describe issues, suggestions, or experiences in a simple text box
    - **Direct Submission** - Send feedback without leaving your development environment
    - **Email Responses** - Receive follow-up communication via email

    **What to Report:**

    - Bugs or unexpected behavior
    - Feature requests or workflow improvements
    - Documentation gaps or unclear instructions
    - Performance issues or errors

    **Impact:** Your feedback directly influences Verdent's development roadmap and helps create a better experience for all users.
  </Tab>
</Tabs>

---

## See Also

<CardGroup cols={3}>
  <Card title="Getting Started" icon="rocket" href="/docs/verdent-for-vscode/getting-started/quick-start">
    Start using Verdent in minutes
  </Card>
  <Card title="Subagent Management" icon="users" href="/docs/verdent-for-vscode/agents-rules/subagent-management">
    Configure specialized sub-agents
  </Card>
</CardGroup>