Milestone 4.5.D.1 — Cross-Track Integration Suite

Status: Planned
Goal: Track D — Exit Gate
Phase: 4.5 — Intelligence Layer
Estimated effort: 3 days
Track: Track D — Exit Gate

Why This Milestone Exists

Phase 1's audit proved individual milestone checkboxes don't guarantee composed behavior works. This suite proves the full pipeline: drifted agent fingerprint → dashboard alert → linked regression run → gated promotion. Uses real Postgres + Redis via testcontainers.

INT-4.5 Test Matrix

ID	Scenario	Assertion
INT-4.5.1	Fresh agent, first 20 sessions	`is_baseline=True` fingerprint, no drift alert
INT-4.5.2	Stable agent, sessions 21–40	Low/absent severity, no alert
INT-4.5.3	Injected drift at session 25	Alert fires within one fingerprinting cycle
INT-4.5.4	Drift alert → regression from suggested-action	`submit-review` triggers against current active directive
INT-4.5.5	2 critical-scenario failures	`promote` rejects with 403
INT-4.5.6	Regression passes, gradual rollout	Advances only if drift/error within threshold
INT-4.5.7	Fingerprinting disabled for org	Zero fingerprint rows written
INT-4.5.8	Auth service outage during fingerprint	Worker retries, no corrupt rows
INT-4.5.9	Cross-tenant: Org A alert invisible to Org B	403 on direct API call
INT-4.5.10	Rollout paused manually	Resumes from paused percentage, not 0

Orientation (indicative)

Named paths, package layouts, libraries, schemas, env vars, and commands anywhere on this page are rough sketches for orientation — inspiration and a baseline, not a required change list.

During implementation, expect to:

open the live tree and follow existing patterns before inventing new ones
research current constraints (latency, tenancy, deploy shape, libraries) more deeply than this page can
advance the design beyond the sketch where measurement or code reality says so
land work in different filenames, merged packages, deferred docs, or new surfaces when the situation calls for it

Prefer outcomes over matching any particular file tree or command sequence.

Areas that may be involved (situational — not a checklist):

Evaluation / benchmarks
Fingerprinting / drift
Directive regression / promotion

Success signals

Outcome-oriented signals that the milestone is in good shape. Exact filenames, package layouts, and commands may differ from any sketches above.

All 10 INT-4.5 cases pass with pytest -m integration, real Postgres + Redis
Suite runtime < 90s in CI
CI job intelligence-integration added to required checks on main

Prerequisites

All Track A–C milestones complete