Start a Team · Lesson 5 of 7

Which tools does YAGNI connect to, and what does it do with them?

Quick answer

YAGNI connects to the tools you already run. The GitHub App comes first and is the only door into your repositories. Linear, Jira, or GitHub Issues carries the tickets a Team works from, Slack carries the decisions and the Daily Brief, and Confluence or Drive takes the export. Every connection is additive: nothing is hosted or migrated.

Which connection comes first?

GitHub, always. It is the first of the two doors on the screen after you sign in, and it installs as an App on your organization, workspace-wide, on the repositories a Team will work in. That installation is the only door into your repositories: every Agent reaches code through it, scoped to the repos a Team attaches, and there is no personal token anywhere in the picture. A member who reaches the door first asks a workspace admin.

Where do the tickets come from?

Linear, Jira, or GitHub Issues. An admin connects the tracker once under Connections, and from then on a Team works from it: your own tickets, matched to you by email, for the my tickets size, or a project, epic, saved view, or repository’s issues attached as the backlog. Add tickets on Work searches the connected trackers. The tracker stays the source of truth: a ticket keeps its key and its assignee on every Work row, and nothing is reassigned or closed until the pull request merges.

Where do the decisions land?

On the case file, and in Slack. When a proposal needs accepting or a plan needs approval, the decision goes to the person it is addressed to, and a reply in the thread lands on the record. There is one ask, rendered wherever you are; answering it in Slack closes it everywhere. Every open decision older than a day reminds its person each morning at 9, in the tray and by email when they have no Slack link. Harper’s Daily Brief posts to the channel the Team names.

What are Confluence and Drive for?

Reading and export. Agents read the docs behind the decisions, and the plans, decisions, and Playbook rules a Team accumulates export back out to Confluence or Drive on request. The knowledge is yours, and the escape hatch is a feature.

Which other tools connect today?

Connector What it feeds
GitHub Repositories, pull requests, issues as a backlog, what merged
Linear and Jira The tickets: yours, a project or epic, what is blocked
Slack Decisions, the Daily Brief, threads that become tickets
Confluence and Notion The docs behind the decisions
Google Drive Docs as supporting context, and the export
Sentry Errors worth proposing a fix for
HubSpot The customer behind the ticket
Stripe Billing truth for billing code
MCP Server, Custom Service, Database Your own tools, spoken to directly

Repository environments live under Connections too: how each repository boots in a sandbox so Wright can build in it and Fletcher can walk a pull request. See repository environment.

What does each Agent read?

Agent Reads
Bailey The backlog the Team attaches, and its Responsibilities
Reeve The plan, the Playbook, and the watch-set you set
Wright The ticket, the repositories, and the decision ledger
Proctor Every pull request on the repos the Team attaches
Fletcher (beta) The pull request revision and the repository environment
Harper The Team’s tickets and their Receipts, plus its channels

Connect only GitHub and a Team still works from tickets you paste or ask @yagni for; each connection makes the same work better grounded.

What never happens?

Nothing is hosted, migrated, or replaced. A workspace-wide connection never attaches a personal source on its own. And credentials never ride in model prompts; they resolve at execution time, inside the tool call, as the security overview describes.

Common questions

Do developers connect their own GitHub accounts?

No. GitHub connects once as an App installation on your organization, workspace-wide, and that installation is the only credential an Agent uses to reach a repository. There is no personal GitHub connection to leak, to expire, or to remove when someone leaves the company.

Do I have to sign in to Linear or Jira myself?

No. Teams always run on the workspace connection an admin made once. You are matched to your tracker account by email, with a one-time Not you? picker if the match is wrong. A personal connection exists only to get a member started before an admin has connected anything, and to sign the CLI's own writes.

What does additive mean in practice?

YAGNI sits on top of your stack. It reads your tools and ships approved work back into them, and it never hosts your apps or migrates your data. Your tickets stay in Linear, Jira, or GitHub Issues, your repos stay on GitHub, your docs stay in Confluence.

What if a tool we run on is not built yet?

Connect it anyway. The MCP Server connector speaks to any MCP server you already run, and the Custom Service and Database connectors cover the rest. Native connectors ship regularly, in the order customers actually run on.

How is connector access secured?

Credentials are customer-scoped, encrypted at rest, resolved only at execution time inside the tool call, and never placed in model prompts. Inference runs on US-hosted providers under zero data retention agreements. The security page documents the full posture.

Read enough. Run it.