Fount is an open-source AI agent runtime for developers and technically curious users who want to build, run and share programmable agents on their own machines. It is built around the Nostr protocol, so agents and people can exchange messages through independent relays rather than through a single company's servers. The program is free and sits in our catalog under Nostr and decentralized social tools.
Most chat front ends let you change a system prompt and a few settings. Fount goes further by letting you customize the agent's logic itself, use community-made plugins and connect characters to editors, browsers, terminals and messaging apps. The trade-off is a steeper learning curve, so you will get the most from it if you are comfortable with code. A default character called ZL-31 helps with setup by handling much of the configuration through conversation.
What Fount Is and Where It Fits
Fount is best understood as a runtime, meaning the environment where agents live, run and talk to tools and people. A regular chat app sends your messages to a model and shows the reply. An agent runtime keeps a character running with its own logic, memory of its task and access to the integrations you connect.
That distinction matters when you choose software. If you only want a faster way to ask questions, a simpler chat front end will do. If you want an agent that keeps a routine, reacts to events and works across several apps, a runtime like Fount is the more appropriate category.
The program also sits in an unusual place between AI tooling and social software. Its chat features come from the same decentralized design as its agents, so the two halves of the project are closely linked rather than bolted together.
How Nostr Relays Shape Decentralized Messaging
Nostr is an open protocol in which people and programs publish signed events to relays, which are servers that store and forward those events. A user does not belong to one platform. Instead, a key pair identifies them, and they can read or publish through any relay that accepts their events.
This design has practical consequences. If one relay goes offline or starts filtering content, you can usually switch to another. Nobody needs permission from a central company to create a new relay, and clients can choose which relays to trust.
For a runtime like Fount, the protocol means agents and people can be reachable without a shared central server. Users still need to think about which relays they connect to, because relay policies differ and affect what they see and who can see them.
Who Benefits Most in Everyday Use
Developers are the clearest audience. A developer who wants an agent to watch a build, answer questions in a terminal and report results to a chat channel can assemble that behavior in one environment instead of stitching together several tools.
Researchers and hobbyists can use shared community agents as working examples. Studying how someone structured an agent's steps is often more useful than reading documentation alone, and adapting an existing agent is faster than starting from zero.
Office users with some technical comfort may find the report export useful for recurring documents. A person who prepares the same weekly summary can set up a character once and then review its drafts, rather than rebuilding the document each time. The learning curve is the main barrier, so less technical users may prefer to let ZL-31 guide the first steps.
Key features
A Standardized Ecosystem Out of the Box
Customizable Agent Logic
Community Agents and Plugins
Quick Report Export
Characters Across Your Tools
Conversational Setup With ZL-31
Peer Chat Without Moderation
Supported coins
Fount's peers normally communicate through Nostr relays. When that path is not available, the runtime can fall back to other peer-to-peer transports: direct connections over a local network, Bluetooth, and WebRTC. The program is listed for Windows, macOS, Linux, Android, Docker and a web or browser extension form, so the same agents can run across several environments.
How to install Fount on Windows
- 1
Open the Official Releases Page
Go to the Releases page of the Fount project on GitHub. Download only from the official website or the GitHub Releases page, and avoid mirrors or download links shared in chats and forums. Keep the page open so you can compare checksums in the next step.
- 2
Choose the Windows Executable
The project describes a direct installation path that uses an exe file from the Releases page, and it is the simplest route on Windows. The most recent release is tagged runner-v0.0.0.6. Download the exe that matches the latest release unless you have a specific reason to use an older build.
- 3
Verify the File
If the release page lists a SHA-256 checksum, calculate the hash of your downloaded file with a tool such as the built-in PowerShell Get-FileHash command and compare the two values. A mismatch means the file should not be run. Scan the file with your antivirus before launching it, and do not disable your protection to get it running.
-
Run the Program and Configure the API
Run the exe and follow the first-launch prompts. Configure the API connection that your agents will use, then start a conversation with the default character ZL-31 to finish the remaining setup. If you prefer containers, the project also lists Docker as an option, but the exe is the most direct path on Windows.
How to use Fount
Once the program is installed and the API is configured, the default character is the quickest place to begin. You describe what you want in plain language, and the character walks through the remaining settings and creates or adjusts agents as needed. Later, you can move on to editing agent logic directly.
- Start the program and confirm that your API connection is configured.
- Open a conversation with the default character ZL-31 and describe the first task you want to automate.
- Ask ZL-31 to create or adjust a character, then review the settings it proposes before accepting them.
- Test the character with a small, low-stakes request so you can see how it responds.
- Connect the character to one additional tool, such as your code editor or terminal, and check that it behaves as expected.
- Browse community agents to compare how others structured their logic, and copy only the parts you understand.
- Export a sample report to confirm the format suits your needs before using the feature for real work.
- Keep notes on what worked, so you can reproduce the setup later or share it with your team.
Pros & cons
- Free to use and open source, so you can read and modify the code
- Runs on several platforms, including Windows, macOS, Linux and Docker
- Customization reaches agent logic, not only prompts and interface settings
- Community agents and plugins give you working examples to study and adapt
- Default character ZL-31 can handle much of the setup through conversation
- Steeper learning curve, and code knowledge is needed for deeper customization
- Some community components may contain malicious code, so you must review them before use
- Windows support is described as likely rather than guaranteed, so confirm the build you download matches your system
- Chat without moderation may not suit communities that need active content rules
Is Fount safe?
Fount is open source, which lets you inspect the code, but that does not make every component safe by default. Download the program only from the official website or the GitHub Releases page of the project, and calculate the SHA-256 checksum of the file before running it. Compare the value with the checksum on the release page when one is published. If no checksum is listed, treat that as a reason for extra caution and scan the file with your antivirus before launching it.
Be especially careful with community agents and plugins. Some may contain malicious code, and they may request access to your API keys, files or messages. Install only components from authors you can check, read their code where possible, and never paste your API keys into a third-party website or a chat window. Keep the program updated, and check the latest version on the official website before installing.
- Official source only. Download from github.com/steve02081504/fount. Ignore ads and lookalike domains.
- Check the SHA-256. Compare the installer hash with the value the developer publishes before running it.
- Never share your seed phrase. No developer, support agent or website — including CryptApps — will ever ask for it.
Alternatives
Most programs in our Nostr category are clients, meaning they focus on reading and publishing notes or direct messages. Amethyst and Gossip fit that pattern, with Gossip being a desktop client written in Rust. Coop concentrates on private messages, while Vega offers a cross-platform desktop experience for general Nostr use.
Fount takes a different approach. It is an agent runtime that uses Nostr for communication, so the focus shifts from browsing feeds to running programmable characters. Command-line tools such as NAK and Nostr Commander RS are closer to Fount in spirit, since they suit users who work in terminals. Nostr VPN, by contrast, uses Nostr signaling for a mesh VPN, which solves networking rather than agent behavior. Choose Fount when agents are the goal and a standard client when reading and writing social content is the goal.
Specifications
- Price
- Free
- Type
- Open source
- License
- —
- Latest version
- runner-v0.0.0.6
- Last update
- 27 Oct 2026
- Windows support
- Likely (builds for Windows; check releases)
- Source code
- github.com/steve02081504/fount
Frequently asked questions
Fount is used to build, run and share programmable AI agents. It lets you customize agent logic, connect characters to tools such as editors, browsers and terminals, and chat with other users over Nostr. It suits developers and technically confident users who want more control than a standard chat front end provides.
Yes, Fount is free and open source. The project's source code is hosted on GitHub, and the program's listing states that there are no expensive subscriptions for its chat features. Check the project page for the most current licensing details before you build on it.
It can be, if you take basic precautions. Download the exe only from the official GitHub Releases page, compare its SHA-256 checksum with the published value when one is listed, and scan it before running. Be careful with community plugins, since some may contain malicious code.
Coding knowledge helps, especially when you want to customize agent logic. The default character ZL-31 can handle much of the setup through conversation, so basic use does not require programming. Expect a learning curve if you plan to build your own agents or plugins from scratch.
Fount users connect through Nostr relays, which store and forward signed events between participants. When relays are not available, the runtime can fall back to peer-to-peer transports such as a local network, Bluetooth or WebRTC. This means you can reach other users without relying on a single central server.
The project's repository shows recent development, and the newest release is tagged runner-v0.0.0.6. Active development does not guarantee stability, so test new builds in a non-critical setup before relying on them for daily work.
Verdict
Fount for Windows suits developers and technically confident users who want programmable agents rather than a simple chat window. Its strengths are customizable logic, community agents and a Nostr-based design that avoids a single central platform. The learning curve is real, and community components need careful review. If you enjoy experimenting with code and want an agent that works across your tools, it is worth a careful trial. If you only need occasional chat assistance, a simpler front end will likely serve you better.