Aegis

Aegis The Proof Phase Specification

This document establishes the 5 Proof Gates for Aegis.

With the architecture and public API contracts frozen, engineering priorities shift from feature building to proving operational reliability, ease of migration, usability, performance, and adoption.

---

1. The 5 Proof Gates

[Proof 1] Operational Reliability (24h/72h/168h Soak Tests & Chaos Scenarios)
      ↓
[Proof 2] Migration Simplicity (Zero-Touch PM2 & Systemd Conversion)
      ↓
[Proof 3] Onboarding Usability (Sub-5 Minute Installation & Detection)
      ↓
[Proof 4] Performance Compliance (Enforced SLO Targets & Zero Regression)
      ↓
[Proof 5] Real-World Adoption (Closed & Public Production Beta Feedback)

---

2. Gate Exit Criteria

Proof 1 — Operational Reliability

Exit Criteria:

• 7-day (168h) continuous soak test completed with zero crashes (validation/soak/72h-report.md).
• All 10 chaos fault-injection scenarios pass cleanly (validation/chaos/chaos-report.md).
• Zero memory leaks (daemon RSS growth < 5%).
• Restart recovery time meets SLO (< 2s).

Proof 2 — Migration Simplicity

Exit Criteria:

• 20 real-world PM2 ecosystem configurations converted without error.
• 10 real-world systemd service units converted without error (validation/migration/migration-report.md).
• Zero critical field mapping failures.

Proof 3 — Onboarding Usability

Exit Criteria:

• New users complete install to first deploy in < 5 minutes.
• Unassisted quickstart success rate > 95%.

Proof 4 — Performance Compliance

Exit Criteria:

• Daemon startup time < 500ms.
• CLI command latency < 100ms.
• Event propagation latency < 10ms.
• All published SLOs consistently met (validation/performance/performance-report.md).

Proof 5 — Real-World Adoption

Exit Criteria:

• 10–20 external production beta deployments verified.
• Documented user feedback collected and addressed.

Proof 6 — System Recoverability

Exit Criteria:

• Verified automated WAL log recovery after power loss (docs/disaster_recovery.md).
• Verified zero-downtime recovery after failed deployment.
• Verified manual fallback path for unmonitored release binaries.