Glossary

Terms you meet when you work with AI agents and with featureloom. Each has one sentence here, and a page of its own with an example and questions.

Agentic codingA way of building software in which an AI agent takes a whole coding task and carries it out in steps on its own, while a person sets the goal and reviews the result.
AI agentA program that pursues a goal on its own over several steps: it decides what to do next, acts through tools and checks the results along the way.
Coding agentAn AI agent, such as Claude Code, Codex or Cursor, that writes and changes a project’s code on its own: it reads code, edits files and runs commands.
EpicA larger body of work that groups the smaller tasks of one topic and shows how many of them are already done.
Human-in-the-loopA way of working with AI in which an AI agent does the task on its own and a person decides at the points that matter for the result.
KanbanA way of managing work in which each task is a card on a board, and the card moves from column to column as the work moves on.
MCP serverA program running on the same computer or at a network address that gives an AI agent tools and data over the Model Context Protocol.
PRD (product requirements document)A document that says what a product or feature must do, for whom and why, before the team starts building it.
Role promptWritten instructions that tell an AI agent what role it plays in the team and how to do its work.
Spec-driven developmentWorking from a specification: a document says what is to be built, an AI agent works from it, and the result is compared with that same document.