ai roadmap
Agents are programs that call tools in a loop. This path treats them as software: schemas, logs, step limits, and least privilege. Multi-agent theatre is optional; reliability is not.
Stages run top to bottom; skills inside a stage can be taken in any order unless a line says otherwise. Locked nodes need their required predecessor first.
Functions as tools, and prompts as specifications.
Function calling, MCP boundaries, and the loop that drives them.
Memory, planning, frameworks compared, and tracing failed runs.
Least privilege, guardrails, and the ethics of delegated action.