juggler-ai
Juggler is a visual workbench for AI coding agents that makes long coding-agent sessions navigable and inspectable. It provides persistent conversation trees, inspectable model transactions and synchronized local or remote sessions through a desktop app and browser.
At a Glance
- Individual software developers
- Professional developers and teams using AI coding agents
- Developers operating remote or headless coding environments
- MCP server developers and people testing/debugging tool integrations
- +1 more
AI Tools by juggler-ai
(2)
Juggler
Visual Workbench for AI Agents
Juggler Code Agent
Open Source AI Coding Agent
Discussions
No discussions yet
Be the first to start a discussion about juggler-ai
Latest News
Juggler v0.6.4 released with macOS multi-display fix, streamed SVG/HTML completion and reduced binned-conversation disk usage.
Juggler v0.6.3 released with whole-working-tree Git review, line/context actions, revised spend accounting and improved Pinboard file pinning.
Juggler v0.6.2 released with conversation-wide spend controls, idle-thread suggestions, expanded syntax highlighting, richer Git details and parallel sub-agents.
Juggler v0.6.1 released with GPT-6 Astra selection, custom OpenAI-compatible endpoints, provider model-limit corrections and conversation paging.
Products & Services
A free, open-source visual workbench for AI coding agents. The native desktop app launches a matching local server; the headless terminal server can run on a dev box or remote machine, with the same live session accessible from the desktop app or a browser.
A public JavaScript extension system for context-item tools, agent-loop strategies, slash commands, info cards, file viewers, Pinboard tabs and UI. The SDK, bundled extensions and examples are Apache-2.0, while the main application is AGPLv3.
Market Position
Juggler positions itself as a visual, inspectable alternative to terminal-first AI coding-agent workflows and conventional chat-style coding tools. Its differentiation is the combination of Miller-column navigation, editable persistent conversation trees, raw model-transaction inspection, local/remote synchronized sessions and an extension architecture in which even built-in tools are plugins; it is aimed at hands-on control rather than a hidden agent loop.
Leadership
Founders
Julian Storer
Developer and founder in audio technology since the 1990s; creator of the JUCE C++ framework, the Tracktion digital audio workstation and the Cmajor programming language. His GitHub profile identifies him as the creator of Juggler and lists London, UK as his location.
Executive Team
Julian Storer
Creator and sole developer
Long-time audio-technology developer and founder; creator of Tracktion, JUCE and Cmajor, and the Juggler repository's primary author.
Founding Story
Julian Storer says he started Juggler after becoming dissatisfied with the existing way of using AI coding agents: editing large blocks of text and absorbing a firehose of information in a terminal. He applied the same pattern behind Tracktion, JUCE and Cmajor—building a tool to address an interface problem he personally encountered—and initially described Juggler as a one-man side project.
Business Model
Revenue Model
Juggler is distributed as a free download and open-source software. It requires no Juggler account; users bring an existing model-provider subscription or their own API keys, so the product does not charge for model usage in the documented offering.
Pricing Tiers
Free to download and use; no signup or Juggler account is required.
Target Markets
- Individual software developers
- Professional developers and teams using AI coding agents
- Developers operating remote or headless coding environments
- MCP server developers and people testing/debugging tool integrations
- Plugin and extension developers
- Developers running long AI-assisted coding sessions
- Inspecting and debugging model context, tool calls, approvals and MCP behavior
- Branching delegated research, tangents or competing implementation approaches without polluting the main thread
- Running an agent where code lives on a workstation, dev box or server while viewing it from multiple devices
- Building, testing and replacing coding-agent tools and workflows through JavaScript extensions
- Working with local models or provider subscriptions through a single interface