KNUCKLE π¦ Β· issue pipeline
βββββββββββββββββββββββββββββββββββββββββββββββββ
βΊ filed needs triage
Β· approved signed off by a maintainer
Β· queued waiting for a contributor to claim
Β· claimed β
Β· done β
βββββββββββββββββββββββββββββββββββββββββββββββββ
domain: β Β· priority: β
size: β Β· next: /approve to queue
---## Finding
The nightly.yml workflow runs 4 critical jobs on a daily schedule:
- nightly-test β 10x race test for flaky detection
- nightly-coverage β per-package coverage gate
- nightly-vuln β govulncheck for new advisories
- nightly-integration β live-network integration tests
None of these jobs have any failure notification mechanism. If a nightly job fails, the only signal is in the GitHub Actions tab β no issue is auto-created, no email/slack is sent, no badge reflects the state.
Risk
Regressions from dependency bumps, upstream API changes, or new vulnerabilities could sit unnoticed for days until someone manually checks the Actions tab.
Recommendation
Add a failure-notification job that runs if: failure() after the test jobs and auto-opens an issue (or pings via workflow dispatch). A simple pattern:
notify-failure:
name: notify on failure
runs-on: ubuntu-latest
needs: [nightly-test, nightly-coverage, nightly-vuln, nightly-integration]
if: failure()
steps:
- uses: actions/checkout@v6
- name: create failure issue
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
gh issue create \
--title "[nightly] Workflow failure $(date +%Y-%m-%d)" \
--body "Nightly workflow failed. See: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}" \
--label "ci,nightly-failure"
Priority
- Impact: high β silent nightly failures = invisible regressions
- Effort: low β ~15 lines of workflow YAML
Filed by quality agent (hold-gated mode)
---## Finding
The
nightly.ymlworkflow runs 4 critical jobs on a daily schedule:None of these jobs have any failure notification mechanism. If a nightly job fails, the only signal is in the GitHub Actions tab β no issue is auto-created, no email/slack is sent, no badge reflects the state.
Risk
Regressions from dependency bumps, upstream API changes, or new vulnerabilities could sit unnoticed for days until someone manually checks the Actions tab.
Recommendation
Add a failure-notification job that runs
if: failure()after the test jobs and auto-opens an issue (or pings via workflow dispatch). A simple pattern:Priority
Filed by quality agent (hold-gated mode)