Raxx · internal docs

internal · gated

CI Notification Posture

Decision date: 2026-05-08 UTC Issue: #1364 Status: Active (pre-launch digest posture)

Summary

Routine CI and cron-job Slack notifications are consolidated into a single daily digest (07:00 UTC via ci-digest-cron.yml). Per-event pings are reserved for production incidents and security findings at CRITICAL/HIGH.

This is a pre-launch posture. At launch — when Raxx has real customers using the platform — routine surfaces flip back to per-event per the flip-back checklist in this document.

Decision

Before Raxx reaches real production users, CI/cron notifications generate noise without actionable signal. Most pings are routine successes or no-drift confirmations. The operator confirmed on 2026-05-08 UTC:

"Digest posture is the standing default until Raxx has real customers."

Condensing to a daily digest reduces per-day routine Slack posts from approximately 12+ (baseline) to 1 (digest) + per-incident pings only.

Digest surface

One Slack message at 07:00 UTC daily, via .github/workflows/ci-digest-cron.yml.

Covers the prior 24 h window:

Section Source
Cron job pass/fail counts GH Actions API (/runs) per workflow file
Flag drift detections drift_run_results DB rows (if CONSOLE_DATABASE_URL set)
Security scan status nightly-security-scan + security-zap run conclusions
PR throughput (merged/open) GH Actions API

Per-event alert surfaces (never digested)

These surfaces post to Slack immediately regardless of digest posture:

Surface Trigger Why
Production incidents H12 / WORKER TIMEOUT / vault degradation tile Customer impact; cannot wait
Security findings CRITICAL or HIGH severity Requires immediate triage
Nightly security scan — failed scan job result = failure Missed scan = operational gap
Nightly security scan — skipped scan job result = skipped/cancelled Missed scan = operational gap
ZAP scan — failed zap-antlers or zap-api result = failure HIGH alerts require immediate review
ZAP scan — skipped (no target) resolve-targets.has_target == false on schedule/dispatch Schedule miss = operational gap
CI (PR) — failure CI workflow conclusion = failure PR gate failure blocks the team
CI Digest build failure build-digest result != success Broken digest = blind spot
Flag drift detected mismatch_count > 0 Unauthorized flag change on prod

Workflows modified by this posture (#1364)

.github/workflows/slack-notify.yml

.github/workflows/synthetic-gate.yml

.github/workflows/nightly-security-scan.yml

.github/workflows/security-zap.yml

.github/workflows/ci-digest-cron.yml (new)

Flip-back checklist (at launch)

When Raxx has real customers, execute the following to restore per-event posture:

Baseline noise levels (pre-#1364)

Source Estimated daily Slack posts
flag-drift-check (every 4 h) ~6
synthetic-gate (every 30 min market hours) ~14 weekdays / ~24 weekends
slack-notify (every PR CI conclusion) ~5–15 (varies by PR activity)
nightly-security-scan (silent on success, no failure alert) 0 (missed gap)
security-zap (no schedule alert) 0 (missed gap)

Total routine Slack posts (pre-#1364, typical weekday): ~25–35

Post-#1364: 1 digest + per-incident pings only.