Terms, filed the same way the documentation is, so a word you meet while building is defined where you would look for it.
Every entry links to the page that explains it properly. This reference orients you; it is not a second source of truth.
The shape of things
A workspace holds projects. One project owns exactly one agent. An agent is built from playbooks and workflows, calls tools, reads a knowledge base, and ships through environments.
Workspace, project and agent: the three containers everything else sits inside.
Playbooks, workflows, steps, tools, and the two prompt layers that shape every turn.
What an agent reads from and writes to: the knowledge base, variables and secrets.
Environments, publishing, merging and traffic splits: how a change reaches real users.
Transcripts, evaluations, tests and credits: reading back what an agent did.
Tokens, traces, the CLI and MCP: the vocabulary for driving Voiceflow over HTTP.
Model, token, RAG, embedding and grounding: industry terms as these docs use them.
Capitalisation carries meaning here. “Agent step” and “Function step” name specific steps; the bare words step and tool stay lowercase.