Historical task source. The Eve task board holds current task state.
This ledger records direct implementation and automated verification evidence for every locally actionable Phase 152 requirement. A checked box is not treated as proof by itself: the corresponding source and focused tests below were inspected and rerun during the completion audit.
Requirement evidence#
| Requirements | Implementation root | Direct evidence |
|---|---|---|
| 152.1.1–152.1.15 | libs/neith/incident-core/ |
Guarded incident lifecycle and required fields; per-severity SLA auto-bump windows; service dependency graph and blast radius; cycle-safe links; audited merge/split; typed custom fields, tags, filters, timeline, impact, closeout, postmortem scaffolding, and anonymising retention. Tests additionally prove whitespace-only lifecycle evidence fails closed and a severity bump starts a new SLA window. |
| 152.2.1–152.2.15 | libs/neith/oncall/ |
Daily/weekly/follow-the-sun rotations, layered schedule resolution, audited overrides, RFC 5545 import, approved swaps, realised time budgets, fairness suggestions, timezone routing, escalation matrices, current/upcoming/API resolution, cross-team handoff, burnout signals, embargoes, and trainee dual-page mode. |
| 152.3.1–152.3.10 | libs/neith/escalation/ |
Validated policy DSL, all-at-once/parallel/round-robin execution, per-position timers, no-ack/reject escalation, conditional/service/role/manager routing, and bounded cycles. Active-recipient authorization prevents spoofed acknowledgements/rejections; parallel mode requires every active recipient to acknowledge. |
| 152.4.1–152.4.15 | libs/neith/alert-ingest/ |
Constant-time HMAC webhook authentication and canonical alert model; native crash/EDR/Nous/synthetic adapters; Prometheus, Grafana, Datadog, New Relic, Splunk, AWS, Azure, GCP, PagerDuty v2, Opsgenie, email, SMS, chat, and mobile parsers. Invalid source, signature, JSON, and parser behavior fails loud. |
| 152.5.1–152.5.15 | libs/neith/notify/ |
Fail-loud injected transports cover push, SMS, voice/DTMF, email, chat, web push, webhooks, bots, IRC, and Matrix; ordered per-severity preferences, DND piercing, provider/channel failover, repeat cadence, provider receipts, acknowledgement reconciliation, and multi-device reachability are executable and tested. |
| 152.6.1–152.6.10 | libs/neith/noise/ |
Alert-key deduplication, time-window grouping, dependency suppression, hysteretic flap detection, maintenance suppression, same-cause clustering, evidence-threshold noise auto-ack, team fatigue, cause ranking, and known-issue attachment. |
| 152.7.1–152.7.15 | libs/neith/war-room/ |
Fail-loud chat/video provisioning, roles and handoff briefing, live chips, runbook/action/decision/agent logs, consent-gated recording, speaker-tagged transcript, stakeholder channel, subincidents, rationale-gated severity changes, and running/final MTTD/MTTR timers. |
| 152.8.1–152.8.15 | libs/neith/status/ |
Branded/custom-domain product model, component and regional rollups, approval-gated broadcasts, 90-day uptime, RSS/Atom/iCal/webhook subscribers, email/SMS subscribers, templates, targeting, maintenance and transition posts, embed payload, translation fallback, and WCAG contrast/content audits. Hierarchy validation fails closed on dangling or cyclic configuration rather than publishing a false-green page. |
| 152.9.1–152.9.15 | libs/neith/runbook/ |
Markdown/command runbooks with typed parameters and git provenance, per-service suggestions, diagnostics, rollback/scale/failover actions, conditional execution, dry-run, fixtures, effectiveness, and grounded draft generation. Approval gates require a non-empty unique approver set and reject unauthorized resume attempts without running effects. |
| 152.10.1–152.10.10 | libs/neith/slo/ |
Availability/latency/quality/correctness and business SLIs, target/window validation, error budgets and burn rate, Google-style multi-window alerts, series/parallel dependency composition, dashboards, review cadence, version history, and release freeze/slowdown policy. |
| 152.11.1–152.11.15 | libs/neith/postmortem/ |
Blameless template/language audit, deduplicated war-room timeline, impact/factors, verified action tracking, read-review-sign workflow, fail-loud wiki publication, external redaction, search/pattern mining, PKM links, retro voting/grouping, cost reporting, and grounded deterministic/LLM executive summaries. |
| 152.12.1–152.12.10 | libs/neith/intel/ |
Milestone-aware MTTD/MTTA/MTTR/MTTF, page/noise/ack/off-hours and on-call metrics, SLA breach and action completion reports, executive trends, DORA feed, team percentiles, drill grades, and checksummed NDJSON warehouse export. |
| 152.13.1–152.13.10 | libs/neith/customer-incident/ |
Manual/automatic ticket attachment, personalized mass replies, VIP severity elevation, tenant impact segmentation, CS/AM outreach, fail-loud conference provisioning, GDPR/HIPAA/CCPA deadlines, legal/comms approval, and gated public trust-center projection. |
| 152.14.1–152.14.10 | libs/neith/gameday/ |
Seeded scheduled drills, guarded/revertible chaos backend, tabletop injects, curated scenarios, weighted rubric, trainee shadowing, risk-ranked failure modes, postmortem-derived scenarios, deterministic synthetic traffic, and ranked team outcomes. |
| 152.15.1–152.15.8 | libs/neith/incident-testkit/ |
Nearest-rank notification P99 gate, provider failover drill, Gini schedule audit, PagerDuty/Opsgenie parser conformance, cache-bust burst simulation, war-room start SLA, and weighted postmortem completeness rubric. |
Automated verification#
The audit ran each project target sequentially because the host has 15 GiB RAM and the repository forbids an unbounded workspace sweep on a small host.
- All 15 Phase 152 Nx test targets pass: 15 files, 256 tests.
- All 15 Phase 152 Nx TypeScript targets pass.
- All 15 Phase 152 lint targets pass; a direct aggregate ESLint audit reports zero errors and zero warnings.
- Focused regression suites pass after the safety corrections:
incident-core25 tests,escalation16 tests,status20 tests, andrunbook17 tests.
The project test targets invoke Vitest through nx:run-commands. This replaces
the deprecated @nx/vite:test executor, which exited before executing tests
with the installed Nx 22.7.4 / Vitest 4.1.7 toolchain.
Deliberately open operational evidence#
- 152.15.9 remains unchecked: proving every Oshun production incident is managed here requires the live production incident stream and organizational rollout.
- 152.15.10 remains unchecked: proving a three-team, six-month external pilot requires customer participation and elapsed field time.
These are the only unchecked Phase 152 requirements and neither can be completed from the repository. All locally actionable implementation and automated validation work is complete.