Docs/Orgs & repos

View as Markdown

Orgs & repos

How Cadence maps your GitHub org and repos to what gets tracked, and who can see what.

Orgs

Cadence tracks a single GitHub org per desktop install, chosen during onboarding. Authentication is GitHub OAuth only - there’s no separate Cadence password. Everything you see - users, teams, repos, sessions - is scoped to that org.

For repo, commit, and PR access, Cadence also uses a GitHub App installed at the org level. That’s what lets it read commit history and git notes, read org/team membership, and post review comments back onto pull requests.

Repos

Repos aren’t added by hand. During onboarding - and any time afterward via Locate - Cadence discovers which of your org’s repos are already cloned on your machine by looking at where your AI coding agents have been running sessions, then cross-references that against the org’s repo list from GitHub.

A repo only has sessions uploaded once it’s enabled (tracked via a local git config flag, ai.cadence.enabled) and it belongs to the org you selected. Nothing uploads from a repo you haven’t enabled.

Teams

Within an org, users can belong to teams (pulled from GitHub team membership). Team-level rollups sit between individual and org views, letting a manager see how their specific team is doing without wading through the whole org.

Access model

Cadence’s access model follows the shape of the org itself:

  • Contributors see their own sessions and coaching in full detail, plus aggregated (not raw) org and team signal.
  • Managers and admins see aggregated org, team, and per-user rollups - cost, adoption, activity, maturity - without needing a raw transcript feed of everyone’s sessions.
  • Anyone can go deeper on a specific session if it’s explicitly shared with them - see Privacy & data.

This is enforced server-side: every query is scoped to your org, and personally-owned data carries an extra layer that only unlocks with an explicit share.