CI failures as a hub status category (provider-namespaced)

[ooloth]

Why

Failing CI builds block other work from landing, but hub has no way to surface them today — you'd need to check each CI provider separately.

Current state

hub status shows PRs and issues only. There is no CI workflow type in the config schema. Multiple CI providers are in use across contexts (GitHub Actions for personal projects, CodeFresh at work), so the design must support both without conflating them.

Ideal state

• CI failures appear in hub status's unified ranked list alongside PRs, issues, and other items — no dedicated CI section
• Each entry shows: repo, workflow/pipeline name, conclusion, age, and URL to the failing run
• The workflow config is provider-namespaced: ci-github-actions and ci-codefresh are distinct hub.toml entries with separate client implementations
• Failures from all configured providers appear together in the unified list, sorted by (urgency, age)
• When all configured runs are passing, no CI items appear

Out of scope

• Re-running failed CI from hub
• Showing passing or queued runs

Starting points

• clients/src/github.rs — pattern for a GitHub-sourced client to follow for ci-github-actions
• config/src/toml.rs — where WorkflowConfig variants are defined
• workflows/src/status.rs — StatusItem enum where new CiFailure variants are added; run() where they're collected and sorted

QA plan

1. Add a ci-github-actions workflow entry to hub.toml for a repo with a failing Actions run
2. Run hub status — expect a CI failure item in the unified list, sorted above lower-urgency items
3. Fix the failing run so it passes, re-run hub status — expect the entry disappears
4. Add a ci-codefresh entry for a repo with a failing Codefresh pipeline, run hub status — expect both providers' failures appear together in the unified list
5. Remove all CI workflow entries from hub.toml — expect no CI items appear

Done when

hub status shows failing CI runs from at least one provider (GitHub Actions or CodeFresh) for configured repos.

[ooloth]

Roadmap update: demoting from next to later.

CI failures are already surfaced via the github-ci workflow, covering the most common case. Adding CodeFresh and other providers is a valid extension but not on the critical path — it's a refinement for a use case that doesn't dominate the daily workflow today.

The more impactful work (urgency ranking, Loki, daemon) should land first. Revisit once the core habit-forming loop is established.