Exposure & Attack Paths
How weaknesses combine into potential routes to business-critical assets, and which remediation breaks the most. As at 31 Oct 2026 · compared with 30 Sept 2026
Attack-path graph
One path at a time — entry point → weakness → privilege or movement → critical asset → inferred business impact
Attack-path register
Every path keeps its source record ID · “No longer observed” is never treated as remediated
Remediation choke points
Ranked by severity × evidence of paths interrupted, critical assets protected and feasibility — not by raw path count
Score = (Σ severity weight × evidence weight of paths interrupted on its own + 1 per business-critical asset no longer reachable) × feasibility. Candidate recommendations are not approved remediation and have no target date. Actions needing a partner action to break a path (e.g. A-1033 with A-1034) show as partial.
What-if remediation simulation
Illustrative remediation scenario — not a verified security outcome.
Selected: none — tick actions in the choke-point table
- None.
None — every critical target is still reachable by a remaining path.
- Each selected action is assumed fully effective and removes exactly the relationships listed for it.
- Only paths in the shared graph are tested; routes not modelled (no attack-path export) are unknown.
- Hypothetical and assessor-reported relationships are treated as present — the result is no stronger than that evidence.
- Nothing here changes risk status, remediation records, historical snapshots or the Cyber Risk Index.
Asset exposure register
Cloud and endpoint assets in one register — 4 Azure VMs matched to their endpoint record by asset ID (never by name); populations that cannot be resolved stay separate
Exposure states are kept separate: known → public endpoint configured → effectively reachable → vulnerable → exploitable → confirmed compromised. “Exploitable” is not assessed for any asset; no compromise is recorded in the incident register.
Identity attack routes
From Identity & Access and Cloud Security — effective privileges are never inferred from role names alone. 23 Azure Owner assignments are assessor-reported. No dedicated Azure RBAC dataset has been ingested or validated.
Shown as identities only — no attack-path relationship is created until a route to them is evidenced.
Endpoint-to-cloud combinations
Candidate scenarios, not confirmed paths — a relationship is shown as observed only where a catalogue dataset supports it
Historical exposure
Paths present at each assessment by evidence status · reduced exposure is shown separately from reduced visibility
No path is observed in source evidence in any assessment — there is no tool attack-path export. Growth from Aug reflects new paths recorded, not necessarily new exposure.
Verified remediation requires a completed register action, recorded dataset completion evidence and an export collected after the change: CAP-1 (A-1051, DS-10 26 May) and IAP-4 (A-1112 on 22 Aug; the same-day DS-03 export cannot show order, so it is verified by the 22 Sep export and shows “no longer observed” in Aug). CAP-5 disappeared from the Aug export without a remediation record, so it is “no longer observed”, not remediated. “Evidence unavailable” marks a visibility gap (e.g. EXP-2 in May before DS-09; IAP-2 in Sep before DS-24) — not reduced exposure. Servers without EDR follow RL-104’s dated history (49 → 31 on 29 Jul).
Management attention
Each item links to its path, remediation record or material risk
RiskLens · Demonstration build · Paths are built from existing synthetic records. “Evidence-backed” refers to simulated evidence, not independent validation.