Map

Software has never been easier to start — or easier to turn into a mess. A simple application becomes a pile of frameworks, dependencies, services, and infrastructure, and AI just makes the pile grow faster. View on GitHub Edit on Replit Free, source included, nothing to sign up for. Most software starts from a blank project. Features get added, dependencies accumulate, architecture evolves, and workarounds become permanent — until changing one thing means worrying about five others. Scalar starts from the opposite premise: a deliberately simple foundation — Python on the server, server-rendered HTML, SQLite, and minimal JavaScript. A working application you can read from top to bottom, understand, and change.

Introduction Why Fraise? Changelog Getting Started Toggle Quickstart Your first memory Core concepts Context engineering Glossary for(let e of document.querySelectorAll(`[data-slot="collapsible-trigger"]`)){let t=e.closest(`[data-slot="collapsible"]`),n=t?.querySelector(`:scope > [data-slot="collapsible-content"]`);t&&n&&(n.style.setProperty(`--radix-collapsible-content-height`,`${n.scrollHeight}px`),e.addEventListener(`click`,()=>{let r=t.dataset.state!==`open`,i=r?`open`:`closed`;if(r&&(n.hidden=!1),n.style.setProperty(`--radix-collapsible-content-height`,`${n.scrollHeight}px`),t.dataset.state=i,e.dataset.state=i,n.dataset.state=i,e.setAttribute(`aria-expanded`,String(r)),!r){let e=()=>{n.dataset.state===`closed`&&(n.hidden=!0)};n.addEventListener(`animationend`,e,{once:!0}),setTimeout(e,350)}}))}document.addEventListener(`click`,e=>{(e.target?.closest(`a[data-sheet-close]`))?.dispatchEvent(new CustomEvent(`sheet-close`,{bubbles:!0}))}); Installation Toggle Homebrew Docker Linux packages Release binaries From source Query Language Toggle Foundations Toggle Design principles Commands Tokens Toggle Keywords Punctuation Identifiers Literals Bounds Parameters Reference Toggle Grammar Warnings Retrieval Toggle Memory Model Toggle Memory graphs Temporality Ranking Toggle Hybrid retrieval Excess transmission Ranking guarantees Retrieval lanes Embeddings Toggle Embeddings overview Producing vectors Dimensions and precision The vector index Architecture Toggle Architecture overview Components Toggle Server Engine Scheduler Database Concurrency Toggle Multi-graph architecture Sched

Open-source, self-hosted database sync with real-time CDC — PostgreSQL, MySQL, SQLite, MongoDB and Redis, any engine to any other. One command to install. Syncle connects to PostgreSQL, MySQL, SQLite, MongoDB and Redis. You pick a table on one of them and where it should go on another. Syncle copies the rows across, then keeps copying as the source changes. Docker with Compose v2 is all you need. Runs on your own machine, MIT licensed. PostgreSQL Syncle MySQL MongoDB Redis One Postgres table feeding three other databases at the same time. A bridge connects one source to one or more destinations. Any engine can sit on either side, so a Postgres table can go to MongoDB, to Redis, or to another Postgres. A destination can also be an HTTP endpoint, if you are feeding a service instead of a database. If the destination table does not exist yet, Syncle creates it and works out the column types for that engine. Writes are upserts, so running a sync twice will not leave you with two copies of a row. A job that dies partway through restarts from the last row it recorded, not from the beginning.