Skip to main content

Agent Extensions

Featured Extensions

Select your coding agent from the menu above to filter extensions. Each extension includes setup instructions and links to official documentation.

Plugins & Skills (8)

Extend your agent with additional capabilities and integrations

Claude Code Plugins

Plugin

Extend Claude Code with pre-built commands, agents, hooks, Skills, and MCP servers. Browse and install from plugin marketplaces.

Capabilities:

  • Custom slash commands
  • Specialized agents for specific tasks
  • Event-driven hooks for workflow automation
  • Model-invoked Skills
  • +1 more
View Setup Instructions
Installation:
  1. Run `/plugin` in Claude Code to browse available plugins
  2. Install a plugin: `/plugin install <plugin-name>@<org>`
  3. Enable/disable plugins: Use plugin management commands
Configuration:
  • Plugins are configured via manifest file (.claude-plugin/plugin.json)
  • Team-level installation via repository .claude/settings.json

Copilot Chat

Plugin

Conversational AI interface in your IDE to explain code, answer questions, debug, and explore solutions with natural language.

Capabilities:

  • Code explanations and debugging
  • Natural language queries about your codebase
  • Test generation with /test command
  • Code fixes with /fix command
  • +2 more
View Setup Instructions
Installation:
  1. VS Code: Install GitHub Copilot Chat extension from marketplace
  2. Or via CLI: code --install-extension GitHub.copilot-chat
  3. Sign in with GitHub account during first launch
Configuration:
  • Access Chat: Ctrl+Shift+A (Windows/Linux) or Cmd+Shift+A (macOS)
  • Use slash commands: /explain, /test, /fix, /doc
  • Reference context with @workspace, @github, @files

Copilot Coding Agent

Plugin

Autonomous agent mode that creates pull requests and writes code based on issues. Assign issues to Copilot for automatic implementation.

Capabilities:

  • Autonomous code generation from issues
  • Automatic pull request creation
  • Responds to review feedback
  • Multi-step task planning and execution
View Setup Instructions
Prerequisites:
  • GitHub Copilot Pro+ subscription ($39/month) or Business/Enterprise plan
Installation:
  1. On GitHub.com: Navigate to an issue
  2. Assign the issue to Copilot (use assignee dropdown)
  3. Copilot creates a branch, analyzes requirements, generates code, and opens a PR
Configuration:
  • Review Copilot's pull request and provide feedback
  • Copilot automatically responds to comments and updates code

Code Review Agent

Plugin

Automated pull request review that analyzes changes, suggests improvements, identifies bugs, and checks for security issues.

Capabilities:

  • Automatic PR analysis
  • Bug detection
  • Security vulnerability scanning
  • Code quality suggestions
  • +1 more
View Setup Instructions
Installation:
  1. Open a pull request on GitHub.com
  2. Click 'Copilot review' button or add comment: @copilot review
  3. Copilot analyzes changes and provides feedback
Configuration:
  • Review suggestions and apply as needed
  • Copilot integrates with existing GitHub PR workflows

Custom Agents

Plugin

Create specialized AI assistants for specific workflows with YAML/JSON configuration. Define system prompts, tool permissions, and knowledge bases.

Capabilities:

  • AWS-specific agents (deployment, Lambda, CDK)
  • Code review agents
  • Testing agents
  • Custom system prompts
  • +1 more
View Setup Instructions
Installation:
  1. Create agent definition file (YAML or JSON)
  2. Use q chat --agent <name> to invoke
  3. Store in .q/ directory for project-specific agents
Configuration:
  • Define agent properties in YAML
  • Set tool permissions (fs_read, fs_write, shell_execute, git_execute)
  • Add knowledge bases and system prompts

Sandboxing & Token Caching

Plugin

Advanced sandboxing for secure command execution and token caching for improved performance.

Capabilities:

  • Secure command execution
  • Token caching
  • Memory management
  • Performance optimization
View Setup Instructions
Installation:
  1. Enable sandboxing in Gemini CLI configuration
  2. Configure token caching settings

Model Selection

Plugin

Choose from Gemini 2.5 Flash, Pro, and other models based on task complexity and pricing needs.

Capabilities:

  • Multiple model tiers
  • Cost optimization
  • Performance tuning
View Setup Instructions
Installation:
  1. Select model in CLI: gemini --model gemini-2.5-flash
  2. Configure default model in settings

Subagent Delegation

Plugin

Delegate tasks to specialized subagents for complex multi-step workflows.

Capabilities:

  • Task delegation
  • Specialized subagents
  • Multi-step coordination
  • Parallel execution
View Setup Instructions
Installation:
  1. Configure subagents in Qwen settings
  2. Qwen automatically delegates tasks to appropriate subagents

MCP Servers (4)

Connect to external tools and services via Model Context Protocol

Model Context Protocol (MCP) Servers

MCP Server

Connect Claude Code to hundreds of external tools and services through MCP servers (GitHub, Slack, Jira, databases, etc.)

Capabilities:

  • GitHub integration
  • Project management (Asana, Atlassian, Linear)
  • Databases (Airtable, HubSpot, PostgreSQL)
  • Design tools (Figma, Canva)
  • +1 more
View Setup Instructions
Installation:
  1. Add MCP server: `claude mcp add --transport http github https://mcp.github.com/mcp`
  2. Add SSE transport: `claude mcp add --transport sse asana https://mcp.asana.com/sse`
  3. List configured servers: `claude mcp list`
Configuration:
  • Project-specific: .mcp.json (checked into source control)
  • Global: ~/.claude/ settings
  • Enterprise: Platform-specific managed files

MCP Integration

MCP Server

Connect to external tools and services via Model Context Protocol for AWS integrations and custom tools.

Capabilities:

  • AWS service integrations
  • Custom tool connections
  • External API access
View Setup Instructions
Installation:
  1. Configure MCP servers: q mcp add --name <name> ...
  2. List servers: q mcp list
  3. Remove servers: q mcp remove --name <name>

MCP Server Integration

MCP Server

Extend Cascade's capabilities with Model Context Protocol servers for GitHub, Figma, databases, and custom tools.

Capabilities:

  • GitHub integration
  • Figma design access
  • Database connections
  • Custom tool integrations
  • +1 more
View Setup Instructions
Installation:
  1. Configure MCP servers in Windsurf settings
  2. Cascade automatically detects and uses MCP tools
Configuration:
  • Use /mcp command to manage servers
  • Enterprise: MCP server whitelisting available

MCP Integration

MCP Server

Connect to external tools via Model Context Protocol for extended functionality.

Capabilities:

  • External tool integrations
  • API access
  • Database connections
View Setup Instructions
Installation:
  1. Configure MCP servers in config.toml
  2. Codex automatically uses available MCP tools

Custom Commands (3)

Define custom workflows and slash commands for common tasks

Workflows

Custom Command

Reusable markdown-based workflows for common tasks like PR review, deployment, testing, and security scans.

Capabilities:

  • Multi-step task automation
  • Plan-driven execution
  • Background or parallel execution
  • Shareable workflow definitions
View Setup Instructions
Installation:
  1. Create workflow markdown file
  2. Define steps and execution plan
  3. Invoke with /workflows command
Configuration:
  • Workflows stored in project configuration
  • Cascade executes workflows with planning agent

AGENTS.md Context

Custom Command

Shareable context file (AGENTS.md) for team workflows, coding standards, and project-specific instructions.

Capabilities:

  • Team-wide context sharing
  • Coding standards documentation
  • Custom workflow definitions
  • Project conventions
View Setup Instructions
Installation:
  1. Create AGENTS.md in repository root
  2. Add team context, workflows, and standards
  3. Codex automatically loads context on startup

Custom Commands (TOML)

Custom Command

Define custom slash commands with TOML format for frequently used workflows and prompts.

Capabilities:

  • Reusable command templates
  • TOML configuration
  • Project-specific workflows
  • 20+ built-in slash commands
View Setup Instructions
Installation:
  1. Create custom command file in TOML format
  2. Place in .qwen/ directory
  3. Use /command-name to invoke
Configuration:
  • TOML format for command definitions
  • Built-in commands: /chat, /memory, /restore, /settings, /stats

Skills (1)

Modular capabilities that your agent can automatically invoke

Agent Skills

Skill

Modular capabilities that Claude autonomously uses based on context. Each Skill consists of a SKILL.md file with instructions plus optional supporting files.

Capabilities:

  • Model-invoked (Claude determines when to use)
  • Personal skills (~/.claude/skills/)
  • Project skills (.claude/skills/)
  • Plugin-bundled skills
View Setup Instructions
Installation:
  1. Create a SKILL.md file with YAML frontmatter (name, description)
  2. Place in ~/.claude/skills/ (personal) or .claude/skills/ (project)
  3. Claude automatically discovers and uses Skills
Configuration:
  • YAML frontmatter: name (lowercase, max 64 chars), description (max 1024 chars)
  • Add supporting files (scripts, templates) in same directory