The repo reads itself
Agent context files are built into the structure. Claude Code opens the folder and immediately understands what each section is for, what module your team is working on, and what to generate next.
The toolkit lives in GitHub. Your team opens it, runs Claude Code, and starts building.
The repo is not a document dump. It is a structured working environment — with policy templates, a prompt library, evaluation rubrics, and agent context files already in place. Open it in Claude Code and the agent reads the structure, understands your city's situation, and helps your team produce a 90% draft policy without starting from scratch.
What your team will find in the repo
The repo is built to work with an agent. Your team lands there with structure, prompts, and context already in place.
The prompts are already written
The prompt library walks your team through every step — policy scoping, risk assessment, draft writing, staff preparation, and council explanation. Run them in order or jump to what you need.
Your team stays in control
The agent generates the draft. Your team reviews, edits, and approves. Every decision that requires legal, labor, or leadership sign-off is flagged explicitly. The policy is yours.
What to open first
Your first session in the repo has four steps. In order:
- Fork the repo to your city's GitHub account at
github.com/agelessextra-design/City-AI-Policy-Toolkit. - Open the folder in Claude Code — or Cursor, Copilot, or another agentic tool that reads file context.
- Read
README.mdfirst — it maps the repo structure and tells you which prompt to run to start your policy work. - Choose your module — start with the area your city most urgently needs: governance, workforce, operations, community trust, or strategic leadership.