AIPM2025-03-11

AIPM prev AI Task Management 2025-02-27

We want to make this web service available to the general public as soon as possible so that we can receive feedback. Think about what we need to do to make this happen, and put it all together in Markdown.

As we plan to offer this web service to the general public, please allow me to review a few things:

1. what is the current development status of this project? (e.g., prototype stage, in internal testing, etc.)
2. do you have a current deployment environment? If so, what infrastructure do you use?
3. are there any particular features or aspects that you would like to emphasize when offering them to the general public?
4. do you have any specific ideas on how to collect user feedback? Based on this information, we will create a more specific development plan.

- 1: Prototype stage, I am the only one using it now. - 2: There is no deployment environment, you have to think about it. when you provide it to others as a web service, you have to think not only about deploying the front end, but also about what to do with the backlog.json that you are reading from fs now. - 3: To be able to try out the whole process with minimal functionality, and to have AI assistance work with a backlog of many tasks that a human can't list. Currently, we are not in that state. It is necessary to verbalize what "one flow" means. - 4: No. We are planning to release it first to communities that already have a way to communicate with us.

Plan Overview:

• Database migration: migration from local JSON files to PostgreSQL, data access layer abstraction implementation
• Authentication system: Implement user authentication and access control using NextAuth.js
• Deployment: front-end in Vercel, database in Supabase, phased release strategy (closed beta -> open beta -> public release)
• Validation strategy: validate data integrity with validate_backlog.py, local workflow testing, PR creation from new branch devin/$(date +%s)-deployment-setup

https://github.com/nishio/ai_project_manager/blob/main/docs/deployment/deployment_plan.md

Discussion: was thinking of putting JSON in a 1 user 1 record firestore. Thinking about pros / cons with the current proposed method. Yes, we will compare and analyze the advantages and disadvantages of using Firestore to store JSON with one user and one record versus using PostgreSQL. We will discuss both approaches in detail.

https://github.com/nishio/ai_project_manager/blob/main/docs/deployment/database_comparison.md

- Database migration: Migration from local JSON files to Firestore (1-user, 1-record system) for multi-user support - Authentication system: Implemented Firebase Authentication to add user registration and login functions and access control - Deployment environment: Vercel for front-end and Firebase for back-end, with phased release (closed beta -> open beta -> public release) - Validation strategy: use validate_backlog.py locally to validate data integrity and test multi-user scenarios. Create PR from new branch (devin/$(date +%s)-deployment-setup)

It would be good. Steps to follow 1: Push the Markdown generated so far 2: I think this plan is too big to do in 10 ACUs. Think about which units could be implemented smoothly and divide them up, write the overall specs and instructions for each individual task in Markdown, and push it.

Noted. First, we will push the Markdown of the plan we have created so far, then we will break the plan down into units that can be implemented in 10 ACUs and create an overall specification and procedures for the individual tasks.

https://github.com/nishio/ai_project_manager/blob/main/docs/deployment/implementation_tasks.md

---

thereafter https://github.com/nishio/ai_project_manager/blob/devin/1741684665-firebase-setup/docs/deployment/iter1.md Work done by humans https://github.com/nishio/ai_project_manager/blob/devin/1741684665-firebase-setup/docs/deployment/human_action_requests.md

I'm up to "4. indexing."

I'm up to "7. deploy."

Wait until deployment is complete (usually within 5 minutes) Access the deployment URL to verify that the application is working properly

You're stuck with a defined but never used type error.

Environment Variable "NEXT_PUBLIC_FIREBASE_API_KEY" references Secret "next_public_firebase_api_key", which does not exist. Ask for a Solution https://chatgpt.com/c/67dbd6d1-e588-8011-92e2-66fce3d8d018

2025-03-24 In vercel.json :

  "env": {
    "NEXT_PUBLIC_FIREBASE_API_KEY": "@next_public_firebase_api_key",

written like ...

It is recommended that sensitive information (e.g. API keys) be stored as Vercel secrets: `vercel secrets add next_public_firebase_api_key "your-api-key"`

2025-03-25 Vercel CLI Overview

https://github.com/nishio/ai_project_manager/pull/67 Deployment was successful, but you can't do anything with no data.

That's nasty.

Text extraction, I didn't tell you how to do it?

In the future, it could be replaced with more sophisticated AI-based extraction logic. For example, it could be extended to use OpenAI's API to extract tasks from more natural sentences. Now it's filled with things that do it the simple way, I see.

---

This page is auto-translated from /nishio/AIPM2025-03-11 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.