chore: Setup squad as a repo assistant

[Stéphane Erbrech (serbrech)]

Description of your changes

This PR sets up Squad for the KubeFleet repository as a trial of human-led AI agent teamwork for issue triage, routing, and contribution support.

Squad adds a persistent team definition to the repo so future Copilot/Squad sessions can reuse shared project context instead of rediscovering KubeFleet each time. The goal is to let Squad help with coordination, issue routing, repetitive triage, and contribution prep while keeping humans accountable for priorities, review, and merge decisions.

Squad composition

The initial team is defined in .squad/team.md and routed by .squad/routing.md:

Member	Role	Primary responsibility
Squad	Coordinator	Routes work, enforces handoffs, and assembles final responses/artifacts.
Ripley	Lead	Architecture, scope control, PR/code review, and higher-level trade-off decisions.
Dallas	Backend Dev	Controllers, reconcilers, rollout, bindings, and work applier changes.
Kane	Backend Dev	Scheduler, plugins, APIs, CRDs, and scheduler snapshots.
Lambert	Tester	Unit/integration/E2E test coverage, Ginkgo/Gomega failures, and validation gaps.
Parker	DevRel	Docs, README/examples, tutorials, and contributor onboarding.
Scribe	Session Logger	Records decisions, session history, and handoff notes.
Ralph	Work Monitor	Watches issues/PR events, triages squad-labeled issues, and routes work to members.
Copilot	Coding Agent	Available for well-scoped implementation issues; auto-assignment is currently disabled.

The team context identifies KubeFleet as a Go/Kubernetes/controller-runtime project, with open GitHub issues from kubefleet-dev/kubefleet as the connected issue source.

Why this is structured this way

Squad has mutable memory: decisions, agent histories, session logs, routing notes, and work-monitor state. That state should not pollute normal KubeFleet feature PRs, so this repo is configured with Squad's two-layer state backend:

• durable Squad state is stored on the dedicated squad-state branch
• commit-scoped notes can capture short-lived context without appearing in PR diffs
• only the static bootstrap/configuration needed to run Squad is committed here

Relevant Squad docs/context:

• Squad README
• State backends
• State Backends - Keep Your PRs Clean
• Watch Monitor

What changed

• Added .github/agents/squad.agent.md so GitHub Copilot can run the Squad coordinator.
• Added .squad/ team configuration, agent charters, routing rules, templates, and skills.
• Configured .squad/config.json with stateBackend: "two-layer".
• Added Squad GitHub workflows for heartbeat, issue assignment, triage, and label sync.
• Added .gitattributes union merge rules for append-only Squad state files.
• Added .gitignore entries for local/runtime Squad state that should not be committed.

How has this code been tested

Not code-path tested. This is repository automation/configuration scaffolding generated by Squad for the KubeFleet trial.

• Associated this change with a known KubeFleet Issue (Bug, Feature, etc). [chore]: Setup Squad as an AI repo assistant #725
• Run make reviewable to ensure this PR is ready for review.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[michaelawyu (michaelawyu)]

Discussed offline ;) Let's merge this and explore a bit more