Outline Driven Development – New AI-Assisted Coding Paradigm; BN
The Problem "Vibing" with LLMs is often too shallow for complex logic, while writing full specifications is cognitively expensive and slow. We need a middle ground that mimics how human programmers gather context—scanning structure before diving into details. The Solution: Outline Driven Development (ODD) I've built a "batteries-included" kit for Gemini/Claude/Codex that uses AST analysis to understand code structure rather than just raw text. This relies on a hyper-optimized Rust toolchain (`ast-grep`, `ripgrep`, `jj`, etc.) to feed precise, structural context to the agent. 1. The…
In plain words
Outline Driven Development is an AI-assisted coding framework that helps developers work more effectively with LLMs like Gemini and Claude. It uses AST analysis to extract structural context from code rather than relying on raw text, striking a balance between casual "vibing" with AI and writing full specifications. Built on optimized Rust tools, it's designed for developers who want better control over how AI understands and generates complex code.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
The Problem "Vibing" with LLMs is often too shallow for complex logic, while writing full specifications is cognitively expensive and slow. We need a middle ground that mimics how human programmers gather context—scanning structure before diving into details. The Solution: Outline Driven Development (ODD) I've built a "batteries-included" kit for Gemini/Claude/Codex that uses AST analysis to understand code structure rather than just raw text. This relies on a hyper-optimized Rust toolchain (`ast-grep`, `ripgrep`, `jj`, etc.) to feed precise, structural context to the agent. 1. The Prerequisites (The Toolchain) Core utilities are installed with `target-cpu=native` for maximum local performance. - Linux/macOS: ``` export RUSTFLAGS="-C target-cpu=native -C opt-level=3 -C codegen-units=1 -C strip=symbols" cargo install lsd ast-grep ripgrep fd-find git-delta tokei && cargo install --locked bat && cargo install --locked --bin jj jj-cli ``` - Windows (PowerShell): ``` $env:RUSTFLAGS="-C target-cpu=native -C opt-level=3 -C codegen-units=1 -C strip=symbols -C link-arg=/LTCG -C link-arg=/OPT:REF" cargo install lsd ast-grep ripgrep fd-find git-delta tokei && cargo install --locked bat && cargo install --locked --bin jj jj-cli ``` 2. Integration Kits I've prepared pre-configured extensions that leverage these local tools for the major AI coding agents. You can install them manually (config injection) or via the CLI wrappers: - Gemini CLI: https://github.com/OutlineDriven/odin-gemini-cli-extension - Claude Code: https://github.com/OutlineDriven/odin-claude-plugin - Codex CLI: https://github.com/OutlineDriven/odin-codex-plugin - Gemini CLI Quick Installation: `gemini extensions install https://github.com/OutlineDriven/odin-gemini-cli-extension` - Claude Code Quick Installation: `claude plugin marketplace add OutlineDriven/odin-claude-plugin && claude plugin install odin-claude-plugin@odin-marketplace`
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, November 2025
the whole month →
- IB
Life & fun · Nov 2025 · bitsnpieces.dev
- BBoing▲782
Life & fun · Nov 2025 · boing.greg.technology