Download Sunder
Install the desktop shell where Agent packages run.
First-party Sunder package family
Build AI workspaces from packages.
Sunder Agent installs into Sunder App. Start with the core package, then add providers, tools, MCP, skills, memory, subagents, execution targets, and Builder capabilities as needed.
Install flow
Install the Agent core, then add the extensions you need.
Agent is intentionally modular. Start with the core runtime package, then add providers, tools, workflow extensions, execution targets, or Builder as your workspace needs grow.
Install Agent
Add sunder.package.agent from the Registry.
Add a provider
Choose OpenAI, Anthropic, Gemini, LM Studio, or future providers.
Add extensions
Install files, shell, web, MCP, skills, memory, subagents, execution, or Builder packages.
Create a workspace
Configure an Agent profile, select models, bind execution, and start a session.
sunder install sunder.package.agentAnswer-ready
Direct answers for AI agent users.
Short, factual definitions for people and answer engines comparing local AI agent workspaces.
What is Sunder Agent?
Sunder Agent is a first-party package family that adds local AI agent profiles, sessions, model providers, workspace tools, MCP integrations, skills, semantic memory, subagents, and execution targets to Sunder App.
How do I install Sunder Agent?
Install Sunder App, open the public Registry, then install the Agent core package. Add only the provider, tool, workflow, and execution packages you need.
sunder install sunder.package.agentCan Sunder Agent use local models?
Yes. The LM Studio provider connects Sunder Agent to local OpenAI-compatible LM Studio endpoints and discovers available local models.
What tools can Sunder Agent use?
Agent tool packages can add file read/write/edit/apply_patch, grep, glob, shell execution, web search, URL fetch, MCP server tools, local execution, and Docker-backed execution.
What is Sunder.Package.Agent.Contracts?
Sunder.Package.Agent.Contracts is a NuGet contracts package used by Agent extensions. It is not itself a Sunder runtime package.
Product proof
Agent runs inside the same Sunder workspace model.
Profiles, sessions, package tools, and workspace context live inside Sunder App. Click the screenshot to inspect the current Agent package experience in more detail.
Package map
Agent packages snap together around the core.
Install only the model providers, tools, workflow extensions, execution targets, and development controls that match the way you work.
Core
Core
sunder.package.agent
Sunder Agent
Adds core sessions, chat, profiles, workspaces, permissions, and orchestration.
Model providers
Provider
sunder.package.agent.provider.openai
OpenAI Provider
Adds OpenAI chat and embedding providers.
Provider
sunder.package.agent.provider.anthropic
Anthropic Provider
Adds Anthropic Claude chat providers.
Provider
sunder.package.agent.provider.gemini
Gemini Provider
Adds Gemini chat and embedding providers.
Provider
sunder.package.agent.provider.lmstudio
LM Studio Provider
Adds local LM Studio chat and embedding providers.
Workspace tools
Tool
sunder.package.agent.tools.files
Workspace Files
Adds file-system tool sources and permission surfaces.
Tool
sunder.package.agent.tools.shell
Workspace Shell
Adds shell command tool sources and permission surfaces.
Tool
sunder.package.agent.tools.web
Web Tools
Adds web fetch and web search tools.
Workflow extensions
Workflow
sunder.package.agent.mcp
MCP Integration
Adds Model Context Protocol server tools to Agent profiles.
Workflow
sunder.package.agent.skills
Skills
Imports reusable SKILL.md folders from local paths or GitHub URLs.
Workflow
sunder.package.agent.subagents
Subagents
Adds specialist child agents and delegated task coordination.
Workflow
sunder.package.agent.memory.semantic
Semantic Memory
Indexes and recalls relevant facts, preferences, and project context.
Execution targets
Execution
sunder.package.agent.execution.local
Local Execution
Adds local machine execution targets for workspace tools.
Execution
sunder.package.agent.execution.docker
Docker Execution
Adds Docker-backed execution targets for isolated workspace runs.
Development controls
Development
sunder.package.agent.builder
Sunder Agent Builder
Adds Sunder package development session controls to Sunder Agent.
Providers
Choose the model surface that fits the task.
OpenAI
API key or ChatGPT Plus/Pro Codex-connected auth, chat, embeddings, streaming, image/PDF input, and native tool calling.
Anthropic
Claude models with streaming, interruptible runs, image/PDF input, and native tool use.
Gemini
Gemini chat and embeddings with image, PDF, audio, and video input support.
LM Studio
Local OpenAI-compatible endpoint support with dynamic local model discovery.
Use cases
From project chat to tool-backed execution.
Sunder Agent is strongest when the AI model can work with installed capabilities instead of being trapped in a single chat box.
- Ask questions about selected project files.
- Read, search, edit, write, and patch workspace files.
- Run non-interactive shell commands through a selected execution target.
- Search the web and fetch pages as markdown, text, or HTML.
- Expose local or remote MCP server tools to Agent profiles.
- Delegate work to configured specialist subagents.
- Recall relevant facts, preferences, project notes, and standing instructions.
- Use local or Docker-backed execution depending on the workspace.
- Use Builder when an Agent workspace should control Sunder package development sessions.
Safety model
Capabilities are explicit packages.
Sunder Agent follows the package model from the source docs: install the core package, then choose which access, provider, execution, and extension surfaces are present.
- File and shell access come from separate tool packages with Agent permission surfaces.
- Provider secrets are stored through Sunder package configuration and secrets abstractions.
- Local and Docker execution are separate packages so users choose what execution surfaces exist.
- Extension packages depend on sunder.package.agent at runtime and are not standalone apps.
Start modular
Install Sunder App, add Agent, then choose your extensions.
Install from the Sunder App marketplace or use the CLI.
Backed by
Agent questions, answered directly.
Is Sunder Agent a separate app?
No. Sunder Agent is a package family installed into Sunder App.
Do I need every Agent extension?
No. Install the Agent core and only the model providers, tools, workflow extensions, execution targets, or development controls you need.
Can Sunder Agent use local models?
Yes. The LM Studio provider connects to a local OpenAI-compatible endpoint and lists local models.
Does Sunder Agent support MCP?
Yes. The MCP extension supports configured MCP servers and exposes MCP tools through Agent profiles.