AI Agent Starter Checklist — AI Builders Network
=================================================

Before you write code:
[ ] Define one user goal the agent must accomplish end-to-end
[ ] List the 1-3 tools needed (no more for v1)
[ ] Choose runtime: framework / SDK / roll your own
[ ] Set up logging for every model turn and tool call

Tool design:
[ ] Each tool does exactly one thing
[ ] Tool descriptions explain when to use, inputs, and output shape
[ ] Tools return structured data, not prose

First vertical slice:
[ ] User input → one tool call → useful response
[ ] Error handling when tool fails
[ ] Demo-ready in under a weekend

Before production:
[ ] Observability dashboard or trace log
[ ] Prompt versioned in git
[ ] Test set of 10+ real inputs
[ ] Cost and latency measured per request