Cointime

Download App
iOS & Android

Frog, Framework for Farcaster Frames

From paradigm by Achal Srinivasan, Georgios Konstantopoulos, awkweb, jxom

Frog is a toolkit for building Frames on Farcaster, a decentralized social network. It lets you create high-performance frames with a few lines of code and deploy it to the infrastructure of your choice. Start by using the documentation or contributing to the code today.

Frog was created as part of the collaboration between Paradigm and Wevm. Our teams see this as a natural extension to the work we're doing building open source tooling like Wagmi, Viem, Reth, Foundry, and others.

We are excited about the future of Frames, a programmable primitive for social apps created by the Farcaster team, and believe it deserves a world-class developer experience.

What are Frames?

Farcaster is a Twitter-like social network, and Frames let developers turn posts ("casts") into interactive elements. It's easy to turn a static post into a dynamic shopping or minting experience in your feed.

Frames turn any cast into an embedded interactive app in a Farcaster client such as Warpcast, redefining how content is viewed, engaged, and shared on social media. Frames are made possible by extending existing patterns defined by the Open Graph protocol, enabling them to be transformed into interactive experiences. With the inception of a new meta paradigm of "interactive social rich objects" to form embedded social user interfaces, the need for a first-class abstraction to enable developers to build high-quality, lightweight, and performant frames without compromising on user and developer experience becomes evident.

What is Frog?

Frog is a framework for Farcaster Frames that enables developers to build high quality, performant, and lightweight frames in just a few blocks of code. Frog comes with a built-in debugger, with live reloading, allowing developers to have a tight feedback loop between their code and frame interface.

See a full demo below.

Integrated on top of Hono, a tiny & fast JavaScript web framework, consumers can use JSX to build frames and integrate it onto any platform or runtime such as: Cloudflare, Vercel, Next.js, Bun, etc.

Essentially, Frog has no vendor lock-in – it can run on any deployment target that supports a JavaScript runtime.

Here is an example “1-file frame”, built on Frog, which you can run via bun run index.ts, and it’ll open a Frame API on localhost:3000, with the debugger accessible under the /dev route. Try it out! It’s that simple, bun install frog.

What makes Frog special?

Devtools are our bread and butter, so we strived to provide a fast, minimal, and feature-rich developer experience.

Here are some reasons why we are excited about in Frog today:

  • Local feature-rich debugger: A local debugger with automatic live reload, build your frames with instant feedback.
  • State Management: A first-class abstraction for state management is achieved through "state derivation" – we can use the deriveState function to derive state for the frame based on state from the previous frame.
  • Connect to Actions: Frog makes it intuitive to connect Frames together with a concept called Actions – instantiate a frame route via app.frame, and then connect it by adding an action path property to your dependent frame.
  • Deploy anywhere: No vendor lock-in – deploy to any JavaScript runtime or framework (Bun, Node.js, Deno, Next.js, Cloudflare Workers, Vercel Edge Runtime, etc).
  • Secure frames: Verification of frame messages is enabled by default provided you supply a Hub API URL (you can also use ours), and ability to sign & verify derived frame state via JSON web signatures.
  • Built on top of Hono: Hono is a super fast and lightweight JavaScript web framework with compelling benchmarks.
  • Build with JSX: Build and visualize frames in code with JSX, a familiar pragma known to React developers.

Why did you build Frog?

We built Frog to 100x the developer experience of writing Farcaster Frames. At a high level, to build a frame, your webpage must provide a set of frame specific meta tags in the document’s block. Akin to Open Graph meta tags, these tags are prefixed with fc:frame:.

To build highly immersive frame applications, with only a set of server-rendered meta tags, presents a unique set of constraints to developers:

  1. There is no concept of client-side reactivity, which makes state management harder than what we’re accustomed to.
  2. There are restrictions on the sizes of fc:frame:* tags.
  3. There are restrictions on the behavior of the fc:frame:* tags (i.e. only 4 buttons, initial state must be empty, etc).
  4. There are security concerns to consider when building frames.
  5. Frame UI (i.e. frame endpoint) and OpenGraph Image UI (i.e. image endpoint) are decoupled.

It can be non-trivial to intuitively build a high-quality & performant frame application in a low-level (vanilla) setting without compromising on these constraints. We built Frog so that we could enable developers to rapidly build frames without thinking about these types of constraints. The framework handles everything under-the-hood so you can build any type of frame you imagine.

We also observed that the local Frame development experience could be improved.

  • The Warpcast Frame Validator is a great tool, especially for validating frames which have been deployed (as the name implies!), however it introduces some challenges for developing frames locally. This is mainly because you can’t insert your locally hosted frame URLs in the validator.
  • Every frame route in Frog comes with a /dev-suffixed endpoint, which consists of the frame devtools – a local debugger for your frames with automatic live-reload. Frog Devtools isn’t intended to replace the Warpcast Frame Validator, but is designed to work alongside it.

Most frames that exist today are culturally built with React & Next.js. This is fine, but we believe that:

  1. You don’t need the entire React API to build a frame – only the JSX pragma.
  2. You shouldn’t have to build multiple Next.js routes/pages to build a frame, nor should you have a sense of vendor lock-in – a frame should exist in a singular module (file), to make it more approachable for contributors, and easy to maintain.

A final observation we have noted is that a lot of frames contain full HTML documents that contain a rich <body> block on top of their frame meta tags. As frames do not utilize the document body at all, this causes a hindrance on frame-load performance. Ideally, frame endpoints should only contain their meta tags – if a user is coming from a browser, we can 302 redirect them to the correct web page accordingly.

What is next?

Frog was built in just three weeks, and Farcaster Frames have existed for about a month. We are still very early in our development journey. We are beyond excited to continue the collaboration between Paradigm and Wevm, and work closely with the Farcaster team on future updates to the Frame specification – ensuring that Frog is closely synced with the latest features. The next thing to be on the lookout for is the new “Transactions” feature for frames added to Frog.

Beyond that, we are excited to:

  1. Pull out the local debugger into a hosted environment for users to play around with any live & hosted frames (whether they are built on Frog or not).
  2. Build Framehouse, a tool to audit your frame for performance, size, and best practices (akin to Lighthouse).

Frog 🐸 is open source (MIT), available for free on Github, along with extensive documentation (built on Vocs!) and a support chatroom. We invite developers to use Frog to build Frames, contribute the codebase, or submit bug reports & feature requests.

If all of this sounds interesting to you, please reach out to: GeorgiosAchalTom, or Jake.

Have ideas on what we should prioritize? Let us know! We’d be thrilled to hear what you are building, or to collaborate on improving Frog and the Farcaster protocol.

Comments

All Comments

Recommended for you

  • A Total of 37,212.18 DMD Permanently Burned Over the Past 7 Days

    July 9, 2026 — According to the latest on-chain data released by DMDAO, a total of 37,212.18 DMD has been permanently burned over the past seven calendar days through the protocol's predefined trading and wealth management burn mechanisms.

  • Whale Transfers 1,133 BTC to Coinbase Prime, Valued at $71.48 Million

    According to Onchain Lens monitoring, a whale transferred 1,133 BTC from Coinbase to Coinbase Prime through an intermediary wallet, valued at $71.48 million.

  • U.S. AI Chip Stocks Decline Before Market Open, Intel Falls Over 3%

    On July 7, U.S. AI chip stocks experienced widespread declines before the market opened. Intel dropped over 3%, while AMD, Qualcomm, and NXP fell more than 2%. TSMC, Broadcom, and Tesla decreased by over 1%, and NVIDIA declined by 0.7%.

  • China's Central Bank Increases Gold Reserves for the 20th Consecutive Month

    As of the end of June, China's gold reserves stood at 75.44 million ounces (approximately 2,346.446 tons), an increase of 480,000 ounces (about 14.93 tons) from the end of May, which reported 74.96 million ounces (approximately 2,331.52 tons). This marks the 20th consecutive month of gold accumulation.

  • China's Foreign Exchange Reserves in June at $341.6262 Billion

    On July 7, China's foreign exchange reserves for June stood at $341.6262 billion, a decrease of $26 billion from the end of May, representing a decline of 0.75%, with expectations set at $343.2 billion.

  • U.S. Storage Stocks Drop Pre-Market, SanDisk and Micron Down Over 4%

    On July 7, U.S. storage concept stocks collectively fell in pre-market trading. Western Digital dropped over 5%, SanDisk and Micron Technology fell over 4%, Seagate Technology declined over 3%, Rambus fell over 2%, and SMI fell over 1%.

  • U.S. Stocks in Optical Communication Sector Drop Pre-Market

    On July 7, stocks in the optical communication sector of the U.S. market collectively fell pre-market. Astera Labs dropped over 4%, while Marvell Technology, Credo Technology, and AXT Inc. fell more than 3%. Tower Semiconductor, MaxLinear, Corning, Applied Optoelectronics, GlobalFoundries, Lumentum, and Qorvo all declined by more than 2%. Coherent, Nokia, Amphenol, and Broadcom dropped over 1%.

  • Pre-market Decline in U.S. Storage Stocks

    In pre-market trading, U.S. storage concept stocks experienced a widespread decline, with Micron Technology falling by 4.8%, SanDisk dropping over 4%, Corning down more than 2%, and Intel decreasing by over 3%.

  • Two Departments: Support for Reinsurance Institutions to Increase Capital and Issue Supplementary Capital Tools

    On July 7, the National Financial Supervision and Administration Bureau and the Shanghai Municipal Government released several measures to accelerate the construction of the Shanghai International Reinsurance Center. Among these measures, they proposed to enhance the quality and efficiency of the reinsurance industry, support reinsurance institutions in increasing capital and expanding shares, and issuing supplementary capital tools to improve the capacity for internal capital accumulation and external capital supplementation, thereby strengthening the reinsurance industry's capabilities. The initiative aims to guide the insurance industry to focus on major national projects, strategic emerging industries, and livelihood security, consolidating insurance and reinsurance underwriting capabilities to enhance risk protection levels. It also supports reinsurance institutions in leveraging their professional technical advantages to assist the insurance industry in reducing risk.

  • Sources: Saudi Arabia Plans to Expand Oil Pipeline to Red Sea, Increasing Capacity by 2 Million Barrels Daily to Bypass Strait of Hormuz

    On July 7, five informed sources revealed that Saudi Arabia is considering expanding the crude oil pipeline capacity to its western coast on the Red Sea, allowing Saudi Arabia and its neighbors to transport more oil without passing through the Strait of Hormuz. This east-west pipeline, built in the early 1980s, has gained strategic importance since the outbreak of the Iran war in February and the disruption of shipping in the Strait of Hormuz. The pipeline can deliver up to 7 million barrels of crude oil per day to the Red Sea port. The CEO of Saudi Aramco stated in May that approximately 2 million barrels are supplied to west coast refineries, while about 5 million barrels are for export. Sources indicate that Saudi Arabia is in preliminary discussions with some neighboring countries regarding the pipeline expansion, aiming to add about 2 million barrels of pipeline capacity per day. It remains unclear whether Aramco's planned expansion involves upgrading existing infrastructure or constructing new pipelines. One source mentioned that the expansion plan also includes a smaller refined oil pipeline. Two sources indicated that the expansion scale could range from 1 million to 2 million barrels per day, with refined oil also being considered. Another source stated that the project would take several years and cost billions of dollars, requiring adjustments to Saudi crude pricing mechanisms.