KOLONİ in 5 minutes¶

KOLONİ's words, each in two or three sentences, with a link to the page that goes deeper. The map shows how they hang together.

KOLONİ's words and how they hang together.

Where you work¶

Project: A folder of yours that agents work in. Each project has its own settings, memory, knowledge and To-do list. To only talk, with no files and no commands, choose No folder · a plain chat. See Quick start.
Chat: One conversation with the Orchestrator, in a project. Each chat has its own permission level, and other folders can join it, read only or read & write. While agents work, a new message waits in a queue. See Chat and canvas.
Turn: Everything that happens for one message: the Orchestrator's answer, the agents it opened, their tools, commands and changes. The canvas draws one turn at a time, from what really happened. See Chat and canvas.

The Orchestrator (Lead): The agent you talk to in a chat. It does small jobs itself and hands separate parts to sub-agents. Behind the scenes its role is named Lead. See Roles.
Sub-agent: An agent that the Orchestrator, or another agent, opens for one part of the work. It changes files in its own working copy, not in your folder. Opening one can ask you first (Settings › This project › Asking and away › Opening a sub-agent). See Permission levels.
Role: What an agent is for: a description, instructions, the tools it may use and, if you like, its own model. KOLONİ comes with roles such as Coder, Reviewer, Tester, Planner and Product owner, and you can add your own on the Agents page. See Roles.
Tier: How strong you say a model is: fast, balanced or strongest, set for each model on the Models page. A role or a job can ask for a tier instead of one model; a model with no tier is only ever picked by name. See Tiers and Agents may use.

Team: A named group of roles, set out in stages: Plan, Build, Check, and Write & keep up. KOLONİ comes with teams such as Web app, Backend / API, Docs and Quality. Teams for this chat on the canvas decides whether the Orchestrator picks teams, you pick them, or no teams are used. See Teams.
Recipe: The drawn steps of a piece of work: who does each step, what runs at once, where it loops back and how it ends. A team runs by a recipe, such as Feature, Bugfix or Evolve. See Recipes and candidates and the Recipe cookbook.
Room: Two to six agents working on one goal at the same time, in one working copy. They share the work out first, each file has one holder, and the room has a budget. Open one with Work together… under More. See Rooms.
Discussion: Several agents talking one question through in rounds, while nobody changes anything. An agent with the discuss tool starts it (the Orchestrator has it); you can join in or end it. See Discussions.

Working copy and Take in: A sub-agent changes files in a copy of your project, outside your folder, so your folder does not change until you say so. Its work comes back on a card: Take in or Discard. A team's recipe keeps its work in one copy until you press Take in. See Working copies and Take in.
Ready for you (cards): A card is a box that waits for your answer, such as "Run this command?" or a file change. Everything that waits is listed in the Ready for you panel (the pill in the top bar) and on the Ready page. The card where a recipe's work ends is Teams' work is ready, with Take in, Fix the findings and Ask for changes. See Ready and approval cards.
Permission levels: What agents may do without asking: Ask before changes, Auto-edit files or Full auto. Each chat has its own. Harmful commands, secret files and files that steer agents ask at every level, with a few exceptions for CLIs run with their own tools (Known limits). See Permission levels.

Memory: What a project remembers between chats: your preferences, the project's rules, facts you approved, and notes agents wrote while they worked. Your preferences and rules reach every agent; a fact reaches an agent whose task names its subject; notes are found only when an agent looks. See Memory.
Knowledge: What agents can look things up in: a docs site, files, a repository or a topic, read once and turned into a short guide. A role uses a knowledge only if you give it to the role. Memory is what the project learned; knowledge is reading you hand to agents. See Knowledge and experts.

To-do: A project's tasks, in your order. Each task runs in a chat of its own, by hand or one after another; agents can propose tasks, and you decide when a task is done. See To-do.
Trigger: Starts a turn by itself: by the clock, when files change, or on a request from outside. Nobody watches that turn, so it may do only what the trigger allows; everything else is refused. For work while you are away, the Orchestrator can also plan a list of jobs. See Triggers and Work while you're away.