--- title: Meet Pi: The Minimalist, Self-Modifying Coding Agent description: Mario, creator of the minimalist coding agent Pi, recounts his path from childhood BASIC programming through libGDX and RoboVM to building Pi as a reaction... canonical: https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Meet Pi: The Minimalist, Self-Modifying Coding Agent | daily.dev og:description: Mario, creator of the minimalist coding agent Pi, recounts his path from childhood BASIC programming through libGDX and RoboVM to building Pi as a reaction... og:url: https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh og:image: https://api.daily.dev/og/posts/mQxPxmedH.png og:image:alt: Meet Pi: The Minimalist, Self-Modifying Coding Agent og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Meet Pi: The Minimalist, Self-Modifying Coding Agent **[Honeypot](https://daily.dev/sources/honeypot)** · 11 min read · 0 upvotes · 0 comments ## Summary Mario, creator of the minimalist coding agent Pi, recounts his path from childhood BASIC programming through libGDX and RoboVM to building Pi as a reaction against the growing complexity and behavioral drift of Claude Code. Pi is built on a simple Node.js/TypeScript stack with four packages (a provider-agnostic AI abstraction, a TUI, an agentic core loop, and the agent itself), and its standout feature is self-modification: the agent can extend or build its own skills. Pi became the core of OpenClaw and reportedly powers a large share of Shopify's agentic coding workloads, ranking ahead of Claude Code on OpenRouter. Mario and his co-founder Armen discuss building a sustainable open, model-agnostic business around Pi while pushing back against 'token maxing' and unsustainable AI velocity, urging developers to stay responsible for code an agent writes. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the Pi coding agent and how is it architected? Pi is a minimalist, open-source coding agent built on a Node.js/TypeScript stack consisting of four packages: an AI abstraction package that normalizes responses across model providers, a TUI (terminal user interface) package, an agentic core package implementing a nested request/tool-call loop, and the coding agent itself combining the three. It notably supports self-modification, letting the agent build its own skills or extensions. _Developers weighing minimalist versus feature-heavy coding agents can follow agent architecture deep dives on daily.dev._ ### Why did Mario stop using Claude Code and build his own coding agent instead? He grew frustrated that Claude Code's team kept adding features at high velocity, which introduced behind-the-scenes changes that made the agent behave inconsistently for the same deterministic workflow. He also disliked depending on a third-party token provider and believed the tool's complexity was unnecessary, favoring a lean codebase of a few thousand lines over a million-line system. _Anyone choosing between agentic coding tools can track these tradeoffs and vendor changes on daily.dev._ ### How widely adopted is the Pi coding agent in production? Shopify reportedly runs about half of its agentic coding workloads on Pi, and on OpenRouter, Pi ranks fourth in usage, ahead of Claude Code and many other open-source agents. Pi also became the agentic core underlying OpenClaw, driving broader indirect adoption. _Teams evaluating agent adoption trends can keep tabs on shifts like this via daily.dev._ ## Similar posts on daily.dev - [Pi: The Minimal Agent Within OpenClaw](https://daily.dev/posts/pi-the-minimal-agent-within-openclaw-tiq1vwnxw) · Armin Ronacher · 7 upvotes · 0 comments --- Tags: [#open-source](https://daily.dev/tags/open-source), [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#typescript](https://daily.dev/tags/typescript) [View this post on daily.dev](https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Meet Pi: The Minimalist, Self-Modifying Coding Agent","url":"https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh"},"datePublished":"2026-08-20T10:19:08.791Z","dateModified":"2026-08-20T12:48:33.365Z","description":"Mario, creator of the minimalist coding agent Pi, recounts his path from childhood BASIC programming through libGDX and RoboVM to building Pi as a reaction...","image":"https://i.ytimg.com/vi/RKHaecOi0CA/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/RKHaecOi0CA/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Honeypot","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Honeypot","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/1cc3a75b98b74698a40f57162b30b841","url":"https://daily.dev/sources/honeypot"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"open-source,llm,ai-agents,typescript","timeRequired":"PT11M","video":{"@type":"VideoObject","name":"Meet Pi: The Minimalist, Self-Modifying Coding Agent","description":"Mario, creator of the minimalist coding agent Pi, recounts his path from childhood BASIC programming through libGDX and RoboVM to building Pi as a reaction...","thumbnailUrl":"https://i.ytimg.com/vi/RKHaecOi0CA/sddefault.jpg","uploadDate":"2026-08-20T10:19:08.791Z","duration":"PT11M","url":"https://api.daily.dev/r/mQxPxmedH","embedUrl":"https://www.youtube.com/embed/RKHaecOi0CA"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Honeypot","item":"https://daily.dev/sources/honeypot"},{"@type":"ListItem","position":3,"name":"Meet Pi: The Minimalist, Self-Modifying Coding Agent"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/meet-pi-the-minimalist-self-modifying-coding-agent-mqxpxmedh#faq","mainEntity":[{"@type":"Question","name":"What is the Pi coding agent and how is it architected?","acceptedAnswer":{"@type":"Answer","text":"Pi is a minimalist, open-source coding agent built on a Node.js/TypeScript stack consisting of four packages: an AI abstraction package that normalizes responses across model providers, a TUI (terminal user interface) package, an agentic core package implementing a nested request/tool-call loop, and the coding agent itself combining the three. It notably supports self-modification, letting the agent build its own skills or extensions. Developers weighing minimalist versus feature-heavy coding agents can follow agent architecture deep dives on daily.dev."}},{"@type":"Question","name":"Why did Mario stop using Claude Code and build his own coding agent instead?","acceptedAnswer":{"@type":"Answer","text":"He grew frustrated that Claude Code's team kept adding features at high velocity, which introduced behind-the-scenes changes that made the agent behave inconsistently for the same deterministic workflow. He also disliked depending on a third-party token provider and believed the tool's complexity was unnecessary, favoring a lean codebase of a few thousand lines over a million-line system. Anyone choosing between agentic coding tools can track these tradeoffs and vendor changes on daily.dev."}},{"@type":"Question","name":"How widely adopted is the Pi coding agent in production?","acceptedAnswer":{"@type":"Answer","text":"Shopify reportedly runs about half of its agentic coding workloads on Pi, and on OpenRouter, Pi ranks fourth in usage, ahead of Claude Code and many other open-source agents. Pi also became the agentic core underlying OpenClaw, driving broader indirect adoption. Teams evaluating agent adoption trends can keep tabs on shifts like this via daily.dev."}}]} ```