Usual provides local memory for Claude Code and Codex, enabling AI agents to learn from past coding conversations and align decisions with user preferences.
Why it matters
GitHub
API Available
overview
Usual is an AI memory tool developed by Paul Jump that enables developers and coders to provide Claude Code and Codex with a memory of past decisions. It learns from previous coding conversations, allowing AI agents to make decisions that align with user preferences. The tool operates locally, eliminating the need for an account or subscription, and is available as an open-source project.
features
Usual is designed to enhance the utility of AI coding assistants by providing a persistent memory layer. Its features focus on local operation and integration with specific AI models.
use cases
Usual is primarily intended for developers and coders who utilize AI assistants like Claude Code and Codex in their workflow. It addresses specific pain points related to AI memory and personalization.
how to use
Usual operates as a local installation, integrating with specified AI coding models to provide memory. Users can get started by accessing its open-source repository.
pricing
Usual is distributed under an open-source license, making it available for free. There are no subscription fees or tiered plans associated with its use.
Similar Tools
Usual occupies a niche in the AI coding assistant landscape by focusing specifically on providing local memory for Claude Code and Codex. Its competitive positioning can be understood by comparing it to broader AI development tools.
Continue integrates directly into your IDE (VS Code, JetBrains) to provide an open-source autopilot for software development, allowing you to chat with any LLM and maintain context across sessions.
While Usual focuses on providing memory specifically for Claude Code and Codex locally, Continue offers a broader integration with various local and remote LLMs directly within your IDE, providing persistent context for your coding workflow. You might give up the explicit focus on Claude/Codex but gain flexibility with other models and a more integrated IDE experience.
Open Interpreter allows local language models to run code (Python, Javascript, Shell, etc.) on your computer, providing a natural language interface to your local machine with persistent session history.
Usual's primary function is to give memory to specific coding models, whereas Open Interpreter focuses on enabling local LLMs to execute code and interact with your system, maintaining a conversational history. You might trade Usual's specialized memory for Claude/Codex for a more general-purpose local AI agent that can execute code and remember past interactions.
GPT-Engineer is a tool that generates an entire codebase based on a prompt, using a conversational approach to refine the requirements and build the project.
Usual focuses on providing memory for ongoing coding conversations with specific AI models, while GPT-Engineer is designed for generating complete projects from scratch through an iterative prompting process. You would be trading Usual's continuous memory for incremental coding for GPT-Engineer's ability to scaffold entire projects based on a high-level description, with its own internal context management for the project generation.
More on Stork
Other tools in this category, matched by shared tags
One short daily email of tools worth shipping. No drip funnel.
one email a day · unsubscribe in two clicks · no third-party tracking
For builders
AI agents read it. Buyers land on it. It answers in eight languages and over MCP. Your tool can have one like it — live in 24 hours.
