ELOWEN DOCUMENTATION

Glossary

Reference on GitHub ↗

Help

Glossary

The words Elowen uses across the CLI, the Web UI, and these docs, in plain language. Each entry links to the page that covers the topic in depth.

Core concepts

Term	Meaning
brain	The conversational agent you chat with — the part that reads your messages, reasons, decides which tools to call, and writes answers. See Brain & Chat.
worker / executor	A non-conversational agent process that runs an assigned task in the background and reports an outcome, instead of chatting with you. See Agents & Providers.
session	One running instance of the brain, bound to a conversation and a set of permissions. The CLI, the web chat, and channels each hold their own sessions. See Brain & Chat.
conversation	The durable message history a session works through. Conversations survive restarts and can be resumed from any surface. See Brain & Chat.
project	A directory tree the agent is allowed to work in. Projects scope file access, shell commands, and search to known roots. See Projects & Workflow.
channel	A chat platform connected to Elowen — Discord, Telegram, Teams, or WhatsApp — where other people can talk to the agent. See Channels.

Automation

Term	Meaning
task	A unit of tracked work on the control plane, with a state, an assignee, and an outcome. Tasks are what workers and missions pick up. See Tasks & Missions.
mission	A standing objective the autopilot works toward by creating and supervising tasks over time, rather than answering a single request. See Tasks & Missions.
autopilot	The mode in which Elowen pursues missions on its own: planning work, assigning it to workers, and checking results without you prompting each step. See Tasks & Missions.
workflow	A directed graph of sub-agents the brain assembles for one request — independent nodes run in parallel, dependents wait for their inputs. See CLI.
sub-agent	A fresh agent with its own clean context, delegated one self-contained task by the brain. It cannot see the parent conversation and can only ever narrow the caller's permissions. See Agents & Providers.
cron job	A prompt Elowen runs on a recurring schedule — daily summaries, periodic checks. See Scheduling.
wake-up	A one-shot scheduled prompt that fires once at a set time or after a delay, then removes itself — used to check back on something that changes over time. See Scheduling.
autonomy level	How much the agent may do without asking first — from read-only up to acting on your behalf — set per account. See Autonomy & Safety.
overseer	The supervisory role that reviews missions and workers: approving plans, checking outcomes, and stepping in when a run goes off track. See Autonomy & Safety.

Extending

Term	Meaning
plugin	A bundle that adds capabilities to Elowen — tools, channels, scheduled jobs — loaded by the daemon. See Plugins.
skill	A markdown instruction file that teaches the agent a repeatable procedure, loaded when a matching task appears. See Skills.
tool	A single callable action the agent can invoke — read a file, run a command, call an API — each with a declared input schema. See Plugins.
MCP	The Model Context Protocol, an open standard for connecting external tool servers. Elowen can consume MCP servers and expose its own tools over MCP. See MCP.

Memory and access

Term	Meaning
memory	Durable facts Elowen stores about you and your projects between conversations — preferences, decisions, environment details — recalled when relevant. See Memory.
embedding	A numeric fingerprint of a piece of text that makes meaning-based search possible — over memories and over your code. See Memory.
role policy	The set of permissions attached to a role — what its holders may read, run, and approve. See Users & Access.
RBAC	Role-based access control: permissions granted through roles assigned to accounts, rather than per-person rules. See Users & Access.
token scope	The slice of access a single API token carries, so an automation or integration gets exactly the permissions it needs and no more. See Users & Access.