Skip to content

Team Configuration

This guide covers the fastest path to mapping repositories and providers to teams.

Org-scoped vs No-org Sync (ClickHouse is the system of record)

As of CHAOS-2600 CS5, ClickHouse is the system of record for the team catalog — both sync paths write ClickHouse directly via insert_teams. There is no Postgres team projection and no Postgres-first control plane. The only difference between the paths is org tagging:

  • Org-scoped (--org ORG): Provider data is written directly to ClickHouse with each team row tagged with org_id (multi-tenant scoping). It does not project to PostgreSQL team_mappings and does not call any Postgres→ClickHouse bridge.
  • No-org (no --org): Provider data is written directly to ClickHouse with no org tag, preserving synthetic/local seeding.

The legacy "Managed Mode" that projected into PostgreSQL team_mappings via TeamDriftSyncService and then bridged into ClickHouse was removed in CS5. The Postgres mapping models/services are dropped in CS6.

Provider Capability Registry

The shared provider capability registry (providers/team_capabilities.py) defines which providers are registered for org-scoped drift discovery. The supported providers are: - GitHub - GitLab - Jira - Linear - Microsoft Teams

The org-scoped CLI path reads this registry to log whether a provider is registered for drift discovery, then writes ClickHouse directly regardless.

Admin team/identity configuration

The admin team and identity surface is ClickHouse-native (CHAOS-2600 CS5): admin team CRUD writes the ClickHouse teams table, and identity→team membership is stored in the ClickHouse identities table with surgical teams.members updates. Admin curation (such as project_keys and repo_patterns) lives in ClickHouse; no Postgres team_mappings / IdentityMapping rows are written. The Postgres-team_mappings drift-review/sync_policy flow is disabled (the drift-review endpoints return HTTP 501). CS6 (CHAOS-2607) deleted the Postgres drift engine; the four 501 stubs remain until CS7 removes them together with the web admin caller (CHAOS-2608), and a ClickHouse-backed rebuild is tracked by CHAOS-2622.

  1. Define your team mapping source (YAML file, Jira, GitHub, GitLab, or synthetic).
  2. Run dev-hops sync teams with the appropriate provider.
  3. Recompute daily metrics so team-scoped views include the latest mappings.

Command reference

For provider-specific commands and flags, use CLI Reference - sync teams.

Next onboarding steps