Best of Elixir — February 2026

  1. 1
    Article
    Avatar of elixirstatusElixirStatus·13w

    Moving my website from NextJS to Phoenix

    A developer migrated their personal website from Next.js to Phoenix/LiveView, driven by MDEx v0.11.0's support for Phoenix HEEX components in markdown. The journey covers using nimble_publisher with a custom MDEx parser, converting React MDX components to Phoenix equivalents, embedding real-time LiveView components in blog posts, clustering multiple Elixir apps on fly.io via Phoenix.PubSub for real-time analytics, replacing Cloudinary with Elixir's image library for banner generation, and building a command bar from scratch. AI coding agents (primarily Amp) were used heavily throughout, with a candid reflection on the trade-offs of AI-assisted development.

  2. 2
    Article
    Avatar of elixirElixir·13w

    Lazy BDDs with eager literal intersections

    José Valim details the latest optimizations to Elixir's set-theoretic type system, specifically the introduction of eager literal intersections on top of the lazy BDD representation introduced in v1.19. Lazy BDDs avoid flattening complex type expressions but can leave redundant nodes when intersections could be resolved immediately. The new optimization eagerly computes intersections between literal nodes in the BDD tree, dramatically reducing tree size and type-checking time—one pathological case dropped from 10 seconds to 25ms. The post covers the mathematical derivation of the optimization, its application to differences, the open vs. closed map trade-off that caused a performance regression in the initial implementation, and how restricting the optimization to closed maps resolved it.

  3. 3
    Article
    Avatar of elixirstatusElixirStatus·12w

    150,000 Lines of Vibe Coded Elixir: The Good, The Bad, and The Ugly

    A founder shares lessons from building 150,000 lines of production Elixir entirely with AI (Claude Code). Elixir's small, terse, immutable nature makes it well-suited for AI coding agents—fewer decisions, fewer tokens, fewer hallucinations. The Tidewave MCP tool extends agent capabilities with live logs and DB access. Key downsides: AI defaults to imperative, defensive Ruby-style code rather than idiomatic Elixir; it cannot debug OTP/async issues or understand Ecto sandbox transaction isolation, causing it to chase red herrings in tests. Architecture decisions still require human oversight. Despite the friction, productivity gains are described as massive.

  4. 4
    Article
    Avatar of elixirstatusElixirStatus·16w

    reposit-bot/reposit: A knowledge-sharing platform where AI agents can contribute solutions, search for similar problems, and improve collectively through voting.

    Reposit is an open-source knowledge-sharing platform built with Elixir/Phoenix that enables AI agents to contribute solutions, search for similar problems using semantic vector search, and collectively improve through voting. The platform uses pgvector for similarity search, OpenAI embeddings, and provides both a web UI and REST API. It includes MCP server integration, Claude Code plugin support, rate limiting, content moderation, and can be self-hosted or used via the hosted service at reposit.bot.

See all Elixir archives