Dexto
A coding agent and general agent harness for building and orchestrating agentic applications.
What is Dexto?
Dexto is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to coding agent and general agent harness for building and orchestrating agentic applications.
A coding agent and general agent harness for building and orchestrating agentic applications.
This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- A coding agent and general agent harness for building and or
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx dextoConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Dexto
Dexto is a full-featured AI agent harness and coding assistant that supports 50+ LLMs (OpenAI, Anthropic, Google, Groq, local Ollama models) and integrates 30+ MCP tools out of the box for tasks ranging from web automation to code generation. It runs as a web UI, a CLI one-shot tool, or as an MCP server itself — making it a flexible orchestration layer for multi-agent pipelines, persistent-memory coding workflows, and human-in-the-loop task automation. Developers use Dexto to build and run complex agentic applications without hand-coding LLM integrations or tool scaffolding.
Prerequisites
- Node.js 18+ and pnpm (for source builds) or curl/PowerShell for the installer script
- At least one LLM API key: ANTHROPIC_API_KEY for Claude, OPENAI_API_KEY for GPT models, or Ollama running locally
- An MCP client if you plan to use Dexto as an MCP server (Claude Desktop, Cursor, etc.)
- Optional: DEXTO_LOG_LEVEL=debug for verbose logging during setup
Install Dexto
Install the Dexto CLI using the one-line installer for macOS/Linux or PowerShell for Windows. Alternatively, build from source using pnpm.
# macOS/Linux/WSL:
curl -fsSL https://dexto.ai/install | bash
# Windows PowerShell:
irm https://dexto.ai/install.ps1 | iex
# From source:
git clone https://github.com/truffle-ai/dexto.git
cd dexto && pnpm install && pnpm install-cliConfigure your LLM provider
Run dexto setup to configure your default LLM provider and model. You can also pass flags directly at runtime to override the model.
dexto setup
# Or run directly with a specific model:
dexto -m claude-sonnet-4-5-20250929 -p "Hello, describe yourself"Create an agent YAML configuration
Define an agent with a specific LLM, system prompt, MCP server connections, and approval policies in a YAML file.
# agent.yaml
llm:
provider: anthropic
model: claude-sonnet-4-5-20250929
apiKey: $ANTHROPIC_API_KEY
systemPrompt: "You are a coding assistant. Only write code that is tested."
mcpServers:
filesystem:
type: stdio
command: npx
args: ['-y', '@modelcontextprotocol/server-filesystem', '.']
permissions:
autoApprove: falseLaunch the Dexto web UI
Run dexto without arguments to start the local web interface where you can manage agents, switch models mid-conversation, and track multi-agent task progress.
dextoRun Dexto as an MCP server
Start Dexto in MCP server mode so other MCP clients can delegate complex agentic tasks to it. The --no-elicitation flag disables interactive prompts for unattended use.
dexto --mode mcp --auto-approve --no-elicitationExecute a one-shot task from the CLI
Use the -p flag for non-interactive one-shot tasks. Add --auto-approve to skip per-tool confirmations for automated pipelines.
dexto -p "Refactor all console.log statements in src/ to use the logger utility" --auto-approveDexto Examples
Client configuration
Claude Desktop configuration to use Dexto as an MCP server, enabling Claude to delegate multi-agent coding tasks to Dexto.
{
"mcpServers": {
"dexto": {
"command": "dexto",
"args": ["--mode", "mcp", "--auto-approve", "--no-elicitation"],
"env": {
"ANTHROPIC_API_KEY": "your-anthropic-key-here",
"DEXTO_LOG_LEVEL": "info"
}
}
}
}Prompts to try
Example prompts to use with Dexto either via its web UI, CLI, or as an MCP tool.
- "Create a React landing page with a hero section and a features grid"
- "Analyze the test coverage of this project and write unit tests for the functions with no coverage"
- "Switch to GPT-4o and continue our conversation"
- "Spawn a sub-agent to search the web for the latest Next.js 15 breaking changes and summarize them"
- "Resume the last session and show me what you were working on"Troubleshooting Dexto
ANTHROPIC_API_KEY not found when starting Dexto
Set the environment variable before running: export ANTHROPIC_API_KEY=sk-ant-... on macOS/Linux or $env:ANTHROPIC_API_KEY='sk-ant-...' in PowerShell. Alternatively configure the key via dexto setup which persists it to ~/.dexto/config.
MCP server connection fails in the agent YAML
Verify the command in mcpServers is on your PATH (e.g., npx is available). Check logs at ~/.dexto/logs/ with DEXTO_LOG_LEVEL=debug for the exact error. Confirm the MCP server package exists and the args are correct.
Dexto web UI does not open after running dexto
Check if port 3000 (default) is already in use with lsof -i :3000. Dexto may output an alternative URL in the terminal — look for the local server address in the startup logs.
Frequently Asked Questions about Dexto
What is Dexto?
Dexto is a Model Context Protocol (MCP) server that coding agent and general agent harness for building and orchestrating agentic applications. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Dexto?
Follow the installation instructions on the Dexto GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Dexto?
Dexto works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Dexto free to use?
Yes, Dexto is open source and available under the NOASSERTION license. You can use it freely in both personal and commercial projects.
Dexto Alternatives — Similar Coding Agents Servers
Looking for alternatives to Dexto? Here are other popular coding agents servers you can use with Claude, Cursor, and VS Code.
Dify
★ 142.2kProduction-ready platform for agentic workflow development.
Ruflo
★ 54.0k🌊 The leading agent orchestration platform for Claude. Deploy intelligent multi-agent swarms, coordinate autonomous workflows, and build conversational AI systems. Features enterprise-grade architecture, self-learning swarm intelligence, RAG integrat
Goose
★ 45.7kan open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
Antigravity Awesome Skills
★ 38.3kInstallable GitHub library of 1,400+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, Antigravity, and more. Includes installer CLI, bundles, workflows, and official/community skill collections.
AgentScope
★ 25.5kBuild and run agents you can see, understand and trust.
Serena
★ 24.5kA coding agent toolkit that provides IDE-like semantic code retrieval and editing tools, enabling LLMs to efficiently navigate and modify codebases using symbol-level operations instead of basic file reading and string replacements.
Browse More Coding Agents MCP Servers
Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Dexto in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Dexto?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.