Skip to content
Agent Fleet

Using Agent Fleet

This shelf answers “how do I do this?” for the person doing the work: starting sessions, following and steering an agent, working with repositories and files, connecting the agent you want to use, and getting unstuck.

If you also manage a team’s members and limits, Administering a tenant is your entry point; if you install and keep a deployment alive, that is Operating a deployment.

Chapters

Section titled “Chapters”

The chapters follow the order you will need them, from your first day to daily work to going further. When you are stuck, it is fine to open 21 Troubleshooting first.

Getting started

Section titled “Getting started”
  1. First day: moved to Start your first session and the pages beside it
  2. Console basics: moved to Find your way around the screen and the pages beside it
  3. Starting a session: kinds, execution method, the Start dialog, the issue tracker

The pages to read first are listed under Getting started.

Daily work

Section titled “Daily work”
  1. Talking to a session: replying, plans, highlights, changed files, reply suggestions, translation
  2. Stopping, resuming and tidying up: stop and resume, handoff, keep awake, locks, cleanup
  3. Repositories and git: cloning, reviewing changes, committing, pushing (SVN too)
  4. Files: the tree, the viewer, editing, Markdown and diagrams
  5. Terminal: copy & paste, round-trip time, font and size
  6. Agents: connecting one, and choosing between them
  7. Chat and memos: questions without a repository, collecting memos

Going further

Section titled “Going further”
  1. Running many sessions at once: the overview, working sets, child sessions, messages, the fleet operator, scheduled runs
  2. Working with others: sharing, handing a session over, the inbox, usage limits
  3. Images: the gallery, generation and the studio
  4. Preview and browser: the browser pane, preview URLs, the development sign-in, a browser the agent opened
  5. External services and signed-in CLIs: gh / bkt / jira / confluence, Discord / Slack, ops tooling (experimental)
  6. Cloud accounts: AWS profiles, SSM hosts, af-aws-exec and Google Cloud
  7. Workspace environment: environment settings, recreating, test databases, a local AWS emulator
  8. Agent memory: what agents remember across kinds and sessions, importing it, and snapshots of their own memory
  9. Using your workspace from outside: MCP, OAuth sign-in, and memory from an external agent

Reference

Section titled “Reference”
  1. Settings: every tab, and “where is that setting?”
  2. Troubleshooting: fixes by symptom, and an FAQ

Also here: Icons, badges and menus for what a mark on the screen means, and the Simple guide, the short version for someone who never opens the terminal (its entry point is Start here if you don’t use the terminal).

What each agent, provider or role can actually do is in the Reference, and the terminology is in the Glossary.

You can open this page in the Console at any time while the workspace is running, from “User guide” in the account menu.