MetaMask Sign-In & Privy Session Fixes for GIGI Projects — April 22, 2026

Today, I began a new authentication strategy for my GIGI Projects by allowing users to sign in with MetaMask, skipping the Privy app. Now, there will be no more session issues for users, and they will be able to log in with ease. I spent 11 hours coding and debugging, and I was able to complete 17 commits; 13 of them were new features and improvements to the site, and 4 of them were crucial fixes.

The focus of my work was the authentication pathway. I was struggling with Privy app’s custom API URL and how it was crossing domains with hashtag.org, which meant that session cookies were going unrecognized. This prevented users from logging in and maintaining their sessions. To solve this issue, I created a direct sign-in route with MetaMask to deal with the bad Privy integration. This was an improvement to the user experience by providing ease of access to GIGI Projects.

While creating this feature, I found some bugs that were associated with Privy’s session management, and one of them was that the app would log users out without informing them if the access token was set to null. This issue led to users receiving an error message when it was really just a lack of access. I was able to solve this issue with an auto wipe for stale browser sessions and an explicit reset call to action. This fix took a lot of debugging to ensure users were guided through the login process instead of running into too many road blocks.

Another bug caused issues with timing of the authentication status. Privy would say that a user was authenticated before the access token was available, which caused the error ‘Could not read Privy session. Try again.’ I fixed this by adding a wait for the access token and then using onComplete and onError callback methods. This change resulted in the authentication flow becoming more reliable and clearer for users.

Outside of authentication, I made great improvements to the management functionality of GIGI Projects. I added a new ideas-to-delivery workflow that enables project owners to complete a structured questionnaire that describes the goals and needs of their project. Automatically, the workflow gets entered once the form is submitted. This workflow has stages for proposal readiness and payment processing via Stripe. This functionality is meant to simplify the project management process and guide users clearly from the inception of an idea to the execution of the project.

I also worked on improving the dashboard by including an overall analytics summary and a GIGI site embed code. The enhanced dashboard has animated cards showing stats related to portal performance, keyword rankings, and competitor research. It was important for me to make an informative and pretty dashboard that provides important information about the users’ projects quickly.

My day included several challenges, especially with the dashboard metrics, integration, and data accuracy. I had to change some code to prevent fallback links from pointing to localhost in production, which may confuse some users. It is important to me that the dashboard looks good and works well. I am trying to create a one-man-show billion dollar company.

Working with Claude Code AI in VS Code is a complete change in this workflow. I have been able to deal with this difficulty without needing to work with a lot of people. It feels good that I can focus on making the agentic web product better without too much work because I can have AI code, debug, and brainstorm.

In short, the progress I made today has brought me more resolve to make GIGI Projects better. I reiterated to myself how crucial testing and user feedback is. I understand there are challenges with working solo, and I know there are challenges. The road is long. Each commit takes me closer to my goals. Each commit brings a clear vision of a successful, scalable platform.