Before you continue
Affected page or file
docs/getting-started/quickstart.md
What is wrong or missing?
The quickstart has no AI-assisted onboarding path. Gas City already has gc prime for teaching agents how to use beads — the same pattern should exist for onboarding itself. Many users discovering Gas City will already be using AI coding agents and would benefit from a copyable prompt block that lets an agent walk them through setup interactively instead of reading docs manually.
What would make this better?
Add a copyable "AI-Assisted Setup" prompt block to the quickstart page. Users paste it into their AI agent (Claude Code, Codex, Cursor, etc.) and get an interactive guided setup:
Suggested prompt block to add:
You are helping me set up Gas City, a multi-agent orchestration SDK.
Read the getting started docs:
Then guide me through setup interactively:
- Check my system for prerequisites (Go 1.25+, tmux, jq) and help me install anything missing
- Install gc via Homebrew or from source — ask which I prefer
- Initialize my first city and explain what just happened
- Help me add my current project as a rig
- Create my first bead and show me how to track work
- Attach me to the mayor session so I can see it in action
- Show me how to prime a coding agent with gc prime
At each step: explain what we're doing and why, run the commands, verify they worked, and only move on when the step succeeds. If something fails, help me debug it.
After setup, suggest next steps based on what I'm building.
This follows the gc prime philosophy — the prompt IS the onboarding. The llms.txt file at docs.gascityhall.com already provides a machine-readable doc index that agents can use.
Additional context
The llms.txt convention (already served at docs.gascityhall.com/llms.txt) makes this easy for agents that support URL fetching. The onboarding prompt could reference it directly so the agent discovers all available docs automatically.
Before you continue
Affected page or file
docs/getting-started/quickstart.md
What is wrong or missing?
The quickstart has no AI-assisted onboarding path. Gas City already has
gc primefor teaching agents how to use beads — the same pattern should exist for onboarding itself. Many users discovering Gas City will already be using AI coding agents and would benefit from a copyable prompt block that lets an agent walk them through setup interactively instead of reading docs manually.What would make this better?
Add a copyable "AI-Assisted Setup" prompt block to the quickstart page. Users paste it into their AI agent (Claude Code, Codex, Cursor, etc.) and get an interactive guided setup:
Suggested prompt block to add:
You are helping me set up Gas City, a multi-agent orchestration SDK.
Read the getting started docs:
Then guide me through setup interactively:
At each step: explain what we're doing and why, run the commands, verify they worked, and only move on when the step succeeds. If something fails, help me debug it.
After setup, suggest next steps based on what I'm building.
This follows the gc prime philosophy — the prompt IS the onboarding. The llms.txt file at docs.gascityhall.com already provides a machine-readable doc index that agents can use.
Additional context
The llms.txt convention (already served at docs.gascityhall.com/llms.txt) makes this easy for agents that support URL fetching. The onboarding prompt could reference it directly so the agent discovers all available docs automatically.