OpenAI builds and deploys advanced AI models like GPT-4o for autonomous agents and workflows.
Usual
0Popularity
About Usual
Usual provides six small, open-source tools designed to help a coding agent reuse past decisions, corrections, and routines stored on the user’s computer. The tools operate locally without requiring an account or uploading data to the website. Users can inspect the evidence, findings, and receipts generated by the tools before deciding whether to apply them. Each tool is scoped to a specific project or set of files, and the system emphasizes transparency by showing source citations, dates, and uncertainty. The tools include Vibecheck for finding useful patterns, Choices for carrying past decisions forward, Recall for retrieving conversations with sources attached, Loops for creating recurring methods, Pop for delivering links in preferred formats, and Escape for helping website visitors leave in-app browsers. The setup is meant to be inspected, adjusted, or removed, and users can share clean recipes of their configurations without exposing private paths or evidence.
Key features
- Local parsing and storage of selected history files
- Scoped analysis of recent projects or specific source files
- Inspectable receipts with source citations and uncertainty details
- Choice review for carrying past decisions into new builds
- Recurring method automation with Loops
- Customizable link delivery with Pop
- In-app browser escape tool with Escape
- Clean recipe sharing without exposing private evidence
Use cases
- Reusing past corrections or patterns in new coding projects
- Automating recurring tasks with scoped local routines
- Retrieving past conversations with attached sources for reference
Pros
- Operates entirely locally without uploading data to the website
- Six focused tools for reusing past work and decisions in coding workflows
- Inspectable receipts and evidence with source citations
- Open-source and MIT-licensed
- No account or history required to start using the tools
Cons
- Public site does not read files or run live scans
- Beta status with limited verification of live agent behavior
- Requires local coding agent setup (e.g., Claude Code or Codex)
- No mobile support for local operations
Frequently asked questions about Usual
What is Usual and who is it designed for?
Usual is a collection of six small, open-source tools designed to help coding agents reuse past decisions, corrections, and routines stored locally on a user’s computer. It is intended for developers or coding agents who want to streamline repetitive tasks by leveraging their own work history without relying on external services or accounts.
How does Usual work with my coding agent?
Usual operates locally by parsing, storing, and searching selected history files on your computer. Your coding agent processes the evidence through its normal model provider, and Usual does not introduce additional model services or telemetry. Users can inspect findings, receipts, and configurations before applying them.
Do I need an account or to upload my data to use Usual?
No account or data upload is required. Usual runs entirely on your computer, and the public website cannot read your files. All parsing, storage, and analysis stay local, ensuring privacy and control over your data.
What are the six tools included in Usual?
The six tools are Vibecheck, Choices, Recall, Loops, Pop, and Escape. Each tool serves a specific purpose, such as finding patterns, carrying past decisions forward, retrieving conversations, creating recurring methods, delivering links in preferred formats, or helping users leave in-app browsers.
Can I share my Usual setup with others?
Yes, you can share clean recipes of your Usual setup, which include tool configurations without exposing private paths or evidence. Recipients can inspect and adapt the setup to their needs without accessing your personal data.
Is Usual open-source, and how can I contribute?
Yes, Usual is open-source under the MIT license. Users can explore the source code, contribute improvements, or report issues on GitHub. Contributions help enhance the tools and make them more useful for the community.
