Quick Start for existing project

[mottych]

I love the idea of an automated AI workflow - thanks for this contribution!

I have been developing a project for a while, comprising four separate repositories: Backend (.net_, AI Backend (python), Frintend (react), admin (react)
I am using gh, gh actions (pulumi for CI)

I do have my own architectural guidelines that I would like the agents to use
I would like to review and approve development plans (Human in the loop step)
I would like all four repositories to work in tandem
I would like the squad team to review and get familiar with the current codebase
I would like to be able to define the workflow so that the system can allocate and align with my current process

My ask:

• Is squad a good fit for this use case?
• What would be a good 'quick start' for an existing project? (high level)

[bradygaster]

I think Squad would be a good fit for your case. You have options - you could start with just a squad, or, you could start with a squad in a folder that contains some markdown files, for example, that document some of your organizational policies. Your initial prompt to set up the team could include instructions for Copilot to read those docs and to factor those docs into the team casting process. Then you could customize your team as needed and have them basically etch all or parts of the policies into their charter and/or the team charter. That would instruct your agents on your organizationl policies and thus, force them to operate with those policies in mind.

I'd also encourage you add another agent whose sole job it is to always validate that all of the policies are being enforced across the entire SDLC your agents follow.

[bradygaster]

Excellent question, @mottyc! Squad is a great fit for your use case.

squad init works on existing projects — it detects your code and proposes a team based on your stack. For your multi-repo setup: document your architectural guidelines as .md files, run squad init in a parent directory, prime the Squad with a prompt referencing your policy docs, and consider a "Policy Validator" team member to audit decisions.

Start with one repo to validate before adding all four. Your setup is totally doable — let us know how it goes! 🚀