Daily BuilderResearch, Roadmap and Plan

[github-actions[bot]]

Repository Overview

KSail is a Kubernetes SDK that bundles common tooling into a single binary. It supports four distributions (Vanilla/Kind, K3s/K3d, Talos, VCluster) across three infrastructure providers (Docker, Hetzner, Omni), with embedded CLI, AI chat TUI, MCP server, and VSCode extension.

As of March 4, 2026, the repository is in excellent health:

• 📊 Open Issues: 3 (all blocked or monitoring-only)
• 📋 Open PRs: 1 (draft gitattributes fix by Copilot bot, chore: exclude *.lock.yaml from GitHub Copilot Review #2768)
• ✅ CI: Stable; golangci-lint clean after comprehensive gosec pass
• 🚀 Velocity: ~30 PRs merged in last 7 days (daily agentic workflows active)

---

🔴 High-Priority Bugs

1. VCluster Image-Pull Retry for GHCR (denied: denied)

Source: Issue #2729 (CI Doctor analysis)
Impact: ❌ CI system tests fail with fetching blob denied for ghcr.io/loft-sh/kubernetes — this is separate from the exit-status-22 / D-Bus transient failures already handled by createWithRetry. The current 7-pattern transientCreateErrors() does NOT cover GHCR image-pull failures occurring mid-provisioning.
Action: Add "fetching blob denied" and "denied: denied" to transientCreateErrors() in pkg/svc/provisioner/cluster/vcluster/provisioner.go, or implement targeted image-pull retry wrapping the Helm install phase.
Effort: Small (1-2 hours)
Files: pkg/svc/provisioner/cluster/vcluster/provisioner.go

2. Monitoring: Transient System Test CI Failures

Source: Issue #2729
Status: 🔒 Blocked — transient infrastructure; PR #2655 (retry increase to 5) was already merged
Action: Continue monitoring; no code change needed unless patterns worsen

---

🚀 High-Impact Roadmap Features

3. Chat TUI Test Coverage (pkg/cli/ui/chat/)

Rationale: The chat TUI has been under active development (PRs #2705, #2700, #2702, #2396). Currently only 3 test files exist (agentmode_test.go, pendingprompt_test.go, yolomode_test.go), covering a small subset of functionality. Key untested areas:

• ChatMode.Next() cycle (Agent → Plan → Agent)
• Model picker interactions (^O)
• Reasoning effort picker (^E)
• Session history (^H)
• Permission handler in Agent mode
• Quota display logic

Effort: Medium (2-4 hours per area)
Files: pkg/cli/ui/chat/model.go, pkg/cli/ui/chat/model_picker.go, pkg/cli/ui/chat/reasoning_picker.go, pkg/cli/ui/chat/permission.go
Pattern: Use export_test.go for black-box testing of unexported methods

4. VCluster Base Image Pre-Warming in CI

Rationale: GHCR denied: denied errors during system tests can be mitigated by pre-pulling ghcr.io/loft-sh/kubernetes:(version) before tests run, similar to the existing retry logic in .github/actions/ksail-system-test/action.yaml.
Effort: Small (1 hour)
Files: .github/actions/ksail-system-test/action.yaml

5. Upstream PR: Update loft-sh/log tablewriter to v1.x API

Rationale: KSail maintains a fork at patches/loft-sh-log/ solely because loft-sh/log/table/table.go uses the incompatible tablewriter v0.0.5 API (issue #2246). Submitting an upstream PR removes the maintenance burden.
Effort: Medium (external contribution, requires fork/PR on loft-sh/log)
Blocked on: External upstream PR acceptance
Action: Fork loft-sh/log, update table/table.go to tablewriter v1.x API, submit PR

6. Upstream Issue: VCluster D-Bus Race Condition

Rationale: Issue #2261 — VCluster SDK install-standalone.sh runs systemctl restart systemd-journald before D-Bus is initialized. KSail has a local workaround (tryDBusRecovery), but filing an upstream issue with loft-sh/vcluster ensures a proper fix.
Effort: Small (30 min to draft and file issue)
Blocked on: Upstream acceptance

---

🔧 Backlog Improvements

7. Update vCluster SDK to Stable Release (#2245)

Status: Currently pinned to v0.32.1 stable (go.mod). Monitor for v0.33.x releases.
Action: Check loft-sh/vcluster releases; upgrade go.mod when new stable available
Effort: Small when released

8. Review and Merge PR #2768 (gitattributes for *.lock.yaml)

Status: Draft PR #2768 adds *.lock.yaml linguist-generated=true to .gitattributes
Action: Review and merge if correct
Effort: Trivial

9. Omni Provider CI Testing (#2547)

Note: PR #2547 (Omni × Docker CI system test) appears to have been merged or superseded; verify status. If still open, review and merge.

10. Add pkg/svc/installer/ package tests

Rationale: Several installer packages have minimal test coverage after the MetalLB tests were added. Extending to cloudproviderkind, cilium, calico, and others would improve quality.
Effort: Medium (2-4 hours per package)

---

🗑️ Stale/Closeable Items

Tracking Issues

• Issue [agentics] Failed runs #2560 [agentics] Failed runs — expires March 5, 2026. Close if no active sub-issues remain.
• Issue [agentics] Failed runs #2376 [agentics] Failed runs — expired Feb 26, 2026. Should be closed.

---

Priority Order for Implementation

Priority	Item	Effort	Impact
1	VCluster GHCR image-pull retry in transientCreateErrors()	Small	High (CI reliability)
2	Chat TUI ChatMode.Next() and model/reasoning picker tests	Medium	Medium (code quality)
3	VCluster base image pre-warming in CI action	Small	Medium (CI reliability)
4	Add tests for more installer packages	Medium	Medium (code quality)
5	File upstream issue for VCluster D-Bus race	Small	Low (upstream)
6	Submit upstream PR for loft-sh/log tablewriter	Large	Low (removes fork)

---

How to Control this Workflow

You can add comments to this discussion to provide feedback or adjust the plan. Available commands:

gh aw disable daily-builder --repo devantler-tech/ksail
gh aw enable daily-builder --repo devantler-tech/ksail
gh aw run daily-builder --repo devantler-tech/ksail --repeat (number-of-repeats)
gh aw logs daily-builder --repo devantler-tech/ksail

---

What Happens Next

The next time this workflow runs, it will begin implementing items from this plan based on priority:

1. High-priority bugs → VCluster GHCR image-pull retry
2. High-impact roadmap features → Chat TUI test coverage, CI pre-warming
3. Backlog improvements → Installer tests, upstream contributions
4. Stale cleanup → Close expired tracking issues

If running in "repeat" mode, the workflow will automatically run again to continue working on items. Humans can review this research and add comments to adjust priorities before the workflow continues.

AI generated by Daily Builder

• expires on Mar 11, 2026, 2:25 PM UTC