Techniques Mapped
| # | Target | STRIDE | Severity | Description | |
|---|---|---|---|---|---|
| 1 | End User | Spoofing | CRITICAL (10.0) | (AI) Credential Stuffing Attack Against OIDC Authentication: Attacker leverages credential reuse from other breaches to gain... | |
| 2 | DevOps Engineer | Elevation of Privilege | CRITICAL (10.0) | (AI) SSH Key Theft and Lateral Movement to CI/CD Server: The DevOps Engineer's SSH session to the CI/CD server is protected... | |
| 3 | CDN Load Balancer | Tampering | CRITICAL (10.0) | Unpatched OS or software vulnerabilities on CDN Load Balancer leading to system compromise | |
| 4 | API Gateway | Tampering | CRITICAL (10.0) | Unpatched OS or software vulnerabilities on API Gateway leading to system compromise | |
| 5 | API Gateway | Tampering | CRITICAL (10.0) | Event injection vulnerability in serverless function API Gateway, leading to arbitrary code execution. |
Threats plotted by Impact (Y-axis, derived from severity score) vs Likelihood (X-axis, derived from risk signals: CVE + CWE + network exposure). Hover over a dot for details.
A higher completeness score improves threat detection accuracy, AI enrichment quality, and severity scoring.
| Check | Score | Wt | Hint |
|---|---|---|---|
| Dataflows authenticated | 0% (0/6) | 15 | Add is_authenticated=true to dataflows that require credentials or token-based auth to reduce false positives in threat scoring. |
| Dataflows encrypted | 83% (5/6) | 15 | Add is_encrypted=true to dataflows that use TLS/SSH/IPSec. Unencrypted flows receive higher STRIDE scores. |
| Dataflows have protocol | 100% (6/6) | 10 | |
| Boundaries trust defined | 100% (3/3) | 15 | |
| Boundaries described | 0% (0/3) | 5 | Add a description to boundaries (e.g. 'DMZ — internet-facing zone'). Descriptions improve AI prompt quality. |
| Servers have classification | 0% (0/6) | 15 | Set classification= (PUBLIC/INTERNAL/RESTRICTED/SECRET) on servers. Classification affects severity scoring. |
| Servers described | 0% (0/6) | 5 | Add description= to servers. Descriptions are used in AI prompts and improve threat specificity. |
| Actors described | 0% (0/2) | 5 | Add description= to actors. Descriptions help the AI identify privilege levels and attack surfaces. |
| Model defines at least one boundary | ✓ | 5 | |
| Model defines at least one dataflow | ✓ | 5 |
The organisation faces a CRITICAL security posture with multiple unmitigated high-severity threats enabling spoofing, tampering, and information disclosure. Critically, goal-driven attack scenarios for authentication bypass via OIDC misconfiguration and CI/CD server pivot to production microservices remain viable after Blue's defences, confirming active exploitability. Immediate investment in multi-factor authentication, patch management, and encryption of database connections is required to reduce existential risk to the business.
| Action | Impact | Effort | Addresses |
|---|---|---|---|
| Enforce multi-factor authentication for all administrative and end-user accounts, and remediate weak authentication on API Gateway routes. | HIGH | MEDIUM | T-0001, T-0008, T-0009, T-0010, T-0011, T-0012 |
| Patch all unpatched OS and software vulnerabilities on CDN Load Balancer, API Gateway, Auth Service, Product Microservice, and CI/CD Server. | HIGH | HIGH | T-0003, T-0004, T-0005, T-0006, T-0013, T-0014, T-0015, T-0016, T-0018 |
| Encrypt all database connections and implement strict access controls to prevent insider threat interception of sensitive data. | HIGH | MEDIUM | T-0007, T-0018, T-0019, T-0020 |
| Deploy input validation and parameterized queries to mitigate SQL injection and event injection vulnerabilities on the Product Database and API Gateway. | HIGH | MEDIUM | T-0017, T-0005 |
| Harden CI/CD server configuration and segment DevOps-to-CICD trust boundaries to block lateral movement pathways. | HIGH | MEDIUM | T-0018, T-0019, GDAF-D6767346, GDAF-77FCFE51 |
Interactive attack-surface map. Click a node to list its threats. Node colour = highest severity. Edge style: solid = encrypted, dashed = plain.
Each row represents a potential pivot chain: an attacker compromises the Entry Point component via its highest-severity threat, then traverses the dataflow to exploit the Target component. Chains are ranked by average severity.
| # | Chain Score | Entry Point | Dataflow / Protocol | Target | Entry Threat | Pivot Threat |
|---|---|---|---|---|---|---|
| 1 | CRITICAL (10.0) | End User |
UserToGateway
HTTPS |
API Gateway |
Spoofing
(AI) Credential Stuffing Attack Against OIDC Authentication: Attacker leverages credential reuse...
|
Tampering
Unpatched OS or software vulnerabilities on API Gateway leading to system compromise
|
| 2 | CRITICAL (10.0) | API Gateway |
GatewayToAuth
HTTPS |
Auth Service |
Tampering
Unpatched OS or software vulnerabilities on API Gateway leading to system compromise
|
Tampering
Unpatched OS or software vulnerabilities on Auth Service leading to system compromise
|
| 3 | CRITICAL (10.0) | API Gateway |
GatewayToProductSvc
HTTPS |
Product Microservice |
Tampering
Unpatched OS or software vulnerabilities on API Gateway leading to system compromise
|
Tampering
Unpatched OS or software vulnerabilities on Product Microservice leading to system compromise
|
| 4 | CRITICAL (10.0) | Product Microservice |
ProductSvcToDB
TCP |
Product Database |
Tampering
Unpatched OS or software vulnerabilities on Product Microservice leading to system compromise
|
Tampering
Unpatched OS or software vulnerabilities on Product Database leading to system compromise
|
| 5 | CRITICAL (10.0) | DevOps Engineer |
DevOpsToCICD
SSH |
CI/CD Server |
Elevation of Privilege
(AI) SSH Key Theft and Lateral Movement to CI/CD Server: The DevOps Engineer's SSH session to...
|
Tampering
Unpatched OS or software vulnerabilities on CI/CD Server leading to system compromise
|
| 6 | CRITICAL (10.0) | CI/CD Server |
CICDToProductSvc
SSH |
Product Microservice |
Tampering
Unpatched OS or software vulnerabilities on CI/CD Server leading to system compromise
|
Tampering
Unpatched OS or software vulnerabilities on Product Microservice leading to system compromise
|
Objective-based attack paths computed by the Goal-Driven Attack Flow engine. Each scenario models a complete adversary journey from entry point to high-value target, with MITRE ATT&CK techniques assigned per hop. Sorted by path score (highest risk first).
| # | Risk | Objective | Actor | Attack Path | Score | Hops | Detection |
|---|---|---|---|---|---|---|---|
|
1
GDAF-B2B36365
|
HIGH |
Authentication Bypass via OIDC Misconfiguration
Attacker exploits a misconfiguration in the OIDC/OAuth2 flow — such as an...
|
External Attacker (Credential Stuffing + OWASP Exploitation)
medium
|
API Gateway → Auth Service | 3.72 🥊 | 2 | 65% |
|
2
GDAF-74EA3363
|
CRITICAL |
Unencrypted Database Connection Interception
Attacker who has gained a foothold within the Private App Zone (e.g. via a...
|
Rogue DevOps Engineer (Insider Threat)
high
|
CI/CD Server → Product Microservice → Product Database | 4.16 🥊 | 3 | 40% |
|
3
GDAF-6EB4C9B3
|
HIGH |
Unencrypted Database Connection Interception
Attacker who has gained a foothold within the Private App Zone (e.g. via a...
|
Ransomware Operator
medium
|
API Gateway → Product Microservice → Product Database | 3.11 | 3 | 60% |
|
4
GDAF-77FCFE51
|
HIGH |
CI/CD Server Pivot to Production Microservice
Attacker compromises the CI/CD server (via RCE, credential theft, or supply...
|
Rogue DevOps Engineer (Insider Threat)
high
|
CI/CD Server → Product Microservice | 3.8 🥊 | 2 | 35% |
|
5
GDAF-D6767346
|
CRITICAL |
CI/CD Server Pivot to Production Microservice
Attacker compromises the CI/CD server (via RCE, credential theft, or supply...
|
Rogue DevOps Engineer (Insider Threat)
high
|
CI/CD Server | 4.23 🥊 | 1 | 20% |
|
6
GDAF-53EFDA43
|
HIGH |
CI/CD Server Pivot to Production Microservice
Attacker compromises the CI/CD server (via RCE, credential theft, or supply...
|
Ransomware Operator
medium
|
API Gateway → Product Microservice | 3.15 🥊 | 2 | 65% |
|
7
GDAF-DD31E3BB
|
HIGH |
API Gateway Bypass to Direct Microservice Access
Attacker discovers that the Product Microservice is reachable directly...
|
External Attacker (Credential Stuffing + OWASP Exploitation)
medium
|
API Gateway → Product Microservice | 3.06 | 2 | 65% |
Unlike GDAF (analyst-defined objectives and actors), these paths are found automatically by following each threat's MITRE ATT&CK tactic progression through the architecture. Best (highest-severity) path shown per STRIDE category.
| # | STRIDE Category | Attack Path | Score | Hops | Threats |
|---|---|---|---|---|---|
| 1 | Elevation of Privilege | CDN Load Balancer → CDN Load Balancer → API Gateway → DevOps Engineer → API Gateway → ProductSvcToDB → ProductSvcToDB → CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database | 79.4 |
CDN Load Balancer
T1195 — Supply Chain Compromise (Initial Access)
CDN Load Balancer
T1072 — Software Deployment Tools (Execution)
API Gateway
T1553 — Subvert Trust Controls (Defense Evasion)
DevOps Engineer
T1082 — System Information Discovery (Discovery)
API Gateway
T1072 — Software Deployment Tools (Execution)
ProductSvcToDB
T1005 — Data from Local System (Collection)
ProductSvcToDB
T1052 — Exfiltration Over Physical Medium (Exfiltration)
CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database
T1498 — Network Denial of Service (Impact)
|
T-0003, T-0003, T-0005, T-0002, T-0004, T-0007, T-0007, T-0073 |
|
Narrative: An attacker begins by compromising the CDN Load Balancer through a supply chain compromise or unpatched software vulnerability, gaining initial access. Using software deployment tools, they execute code on the load balancer and pivot to the API Gateway, where they subvert trust controls via an event injection vulnerability to achieve arbitrary code execution. The attacker then discovers system information through the DevOps Engineer's SSH session and leverages software deployment tools again through the API Gateway to move deeper. Credentials and data are collected from the local system as they travel across the unencrypted channel from the Product Microservice to the Product Database, and exfiltrated over the physical medium. Finally, the attacker exploits the SSH dataflows between the CI/CD server and the Product Microservice to inject malicious code or backdoors, leveraging trusted access for impact.
Business impact: Successful exploitation leads to full compromise of the Product Database and microservice, enabling data exfiltration, injection of malicious code or backdoors via the CI/CD pipeline, and potential lateral movement into broader infrastructure, resulting in significant reputational damage, regulatory fines, and operational disruption.
|
|||||
| 2 | Information Disclosure | CDN Load Balancer → CDN Load Balancer → API Gateway → DevOps Engineer → API Gateway → ProductSvcToDB → ProductSvcToDB → CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database | 79.4 |
CDN Load Balancer
T1195 — Supply Chain Compromise (Initial Access)
CDN Load Balancer
T1072 — Software Deployment Tools (Execution)
API Gateway
T1553 — Subvert Trust Controls (Defense Evasion)
DevOps Engineer
T1082 — System Information Discovery (Discovery)
API Gateway
T1072 — Software Deployment Tools (Execution)
ProductSvcToDB
T1005 — Data from Local System (Collection)
ProductSvcToDB
T1052 — Exfiltration Over Physical Medium (Exfiltration)
CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database
T1498 — Network Denial of Service (Impact)
|
T-0003, T-0003, T-0005, T-0002, T-0004, T-0007, T-0007, T-0073 |
|
Narrative: An attacker begins by compromising the CDN Load Balancer through a supply chain compromise or unpatched software vulnerability, gaining initial access. Using software deployment tools for execution, the attacker pivots to the API Gateway, where they subvert trust controls via an event injection vulnerability to achieve arbitrary code execution. With engineer credentials obtained through SSH key theft, the attacker navigates to the CI/CD server, then leverages trusted SSH dataflows to inject malicious code into the product microservice. Finally, exploiting unencrypted credential transmission, the attacker queries the product database and exfiltrates data over the physical medium.
Business impact: Successful exploitation leads to full compromise of the product microservice and database, resulting in unauthorized data exfiltration, potential injection of malicious code or backdoors via the CI/CD pipeline, and significant reputational and financial damage from the breach of sensitive product information.
|
|||||
| 3 | Tampering | CDN Load Balancer → CDN Load Balancer → API Gateway → DevOps Engineer → API Gateway → ProductSvcToDB → ProductSvcToDB → CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database | 79.4 |
CDN Load Balancer
T1195 — Supply Chain Compromise (Initial Access)
CDN Load Balancer
T1072 — Software Deployment Tools (Execution)
API Gateway
T1553 — Subvert Trust Controls (Defense Evasion)
DevOps Engineer
T1082 — System Information Discovery (Discovery)
API Gateway
T1072 — Software Deployment Tools (Execution)
ProductSvcToDB
T1005 — Data from Local System (Collection)
ProductSvcToDB
T1052 — Exfiltration Over Physical Medium (Exfiltration)
CI/CD Server → DevOpsToCICD → CICDToProductSvc → Product Microservice → Product Database
T1498 — Network Denial of Service (Impact)
|
T-0003, T-0003, T-0005, T-0002, T-0004, T-0007, T-0007, T-0073 |
|
Narrative: An attacker begins by compromising the CDN Load Balancer through a supply chain compromise or unpatched software vulnerability, gaining initial access. Using software deployment tools, they execute code on the load balancer and pivot to the API Gateway, where they subvert trust controls via an event injection vulnerability to achieve arbitrary code execution. The attacker then discovers system information through the DevOps Engineer's SSH session, leveraging weak two-factor authentication and IP filtering protections. From the API Gateway, they exploit unpatched software vulnerabilities to move laterally, exfiltrating credentials sent over unencrypted channels from the Product Microservice to the Product Database. Finally, the attacker leverages SSH access through the CI/CD pipeline to inject malicious code or backdoors, ultimately compromising the Product Database.
Business impact: Successful exploitation leads to full compromise of the Product Database, enabling data exfiltration, persistent backdoors, and potential disruption of product availability. The attacker can inject malicious code via the CI/CD pipeline, risking supply chain contamination that affects all downstream deployments and erodes customer trust in the product's integrity and security.
|
|||||
Each top attack path below was challenged in a multi-round Red (attack) vs. Blue
(defence) exchange. Read it for what Blue blocked, where detection gaps remain, and
what Red tried and failed. The verdict is a qualitative outcome — the numeric score
adjustment it drives is directional, not precise (see docs/evaluation.md).
Verdict: Viable but detected — Blue has reliable coverage. Red tried and failed to bypass 2 technique(s); 2 detection gap(s) remain.
| Step | Control family | Covered | Detail | Confidence |
|---|---|---|---|---|
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
docs/evaluation.md)
Verdict: Viable but detected — Blue has reliable coverage. Red tried and failed to bypass 2 technique(s); 2 detection gap(s) remain.
| Step | Control family | Covered | Detail | Confidence |
|---|---|---|---|---|
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
| T1195 event log clearance on CI/CD server | SIEM | ❌ | No SIEM correlation rule for T1195 audit log clearance events on CI/CD server | medium |
docs/evaluation.md)
Verdict: Viable but detected — Blue has reliable coverage. Red tried and failed to bypass 2 technique(s); 2 detection gap(s) remain.
| Step | Control family | Covered | Detail | Confidence |
|---|---|---|---|---|
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
docs/evaluation.md)
Verdict: Viable but detected — Blue has reliable coverage. Red tried and failed to bypass 2 technique(s); 2 detection gap(s) remain.
| Step | Control family | Covered | Detail | Confidence |
|---|---|---|---|---|
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
docs/evaluation.md)
Verdict: Viable but detected — Blue has reliable coverage. Red tried and failed to bypass 1 technique(s); 1 detection gap(s) remain.
| Step | Control family | Covered | Detail | Confidence |
|---|---|---|---|---|
| hop 2 SSH access | SIEM | ❌ | No correlation rule for password-auth SSH from external IP ranges | medium |
docs/evaluation.md)
Threat severity is calculated on a scale of 1.0 to 10.0 using the following factors: 1. **Base Score**: Each STRIDE threat category has a predefined base score (ElevationOfPrivilege: 9.0, Tampering: 8.0, InformationDisclosure: 7.5, Spoofing: 7.0, DenialOfService: 6.0, Repudiation: 5.0). 2. **Impact and Likelihood**: If provided (scale 1–5), their product is normalised and added. 3. **Target Multipliers**: Per-element multipliers from the '## Severity Multipliers' section of the model. 4. **Protocol Adjustments**: SSH +0.5, HTTPS -0.3, HTTP +0.2. 5. **Data Classification**: PUBLIC ×1.0, RESTRICTED ×1.2, SECRET ×1.5, TOP_SECRET ×2.0 multiplier. 6. **VOC Context (when available)**: - Known CVE match for this target: +0.5 - High-risk CWE class (injection, buffer overflow, hardcoded creds…): +0.3 - Network-exposed without authentication or encryption: +0.7 - Active D3FEND defensive controls in place: -0.5 The final score is clamped to [1.0, 10.0] and mapped to INFORMATIONAL / LOW / MEDIUM / HIGH / CRITICAL.
Legend:
* : Indicates a CAPEC or MITRE mapping was manually added to improve relevance.
: Indicates a mitigation has been implemented (as defined in the implemented mitigations file).
| # | Target | Business Value | Source | Description | Conf. | CVEs | Severity | Risk Signals | CAPEC | MITRE Techniques | MITRE Mitigations | D3FEND Mitigations | CIS Mitigations | NIST Mitigations | OWASP Mitigations | Detection (SOC) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 |
Product Database | N/A |
pytm
|
Denial of Service against Product Database through resource-intensive queries or excessive connections
TK-86389529
|
1.00
|
|
HIGH (8.9) |
D3F
|
|
|
|
—
|
||||
2 |
API Gateway | N/A |
pytm
|
Improper rate limiting on API Gateway API Gateway, leading to Denial of Service.
TK-594625FE
|
1.00
|
|
HIGH (8.4) |
NET
D3F
|
—
|
|||||||
3 |
Product Database | N/A |
AI
|
(AI) Database Service Exhaustion via Sustained Connection Flood: The Product Database is not protected by a WAF or IPS and relies on redundant infrastructure only. An attacker who discovers the database endpoint IP can send a high volume of connection requests, exhausting database connection pools and CPU resources, leading to availability degradation for the Product Microservice and downstream customers.
TK-F64DC492
|
0.90
|
|
HIGH (8.9) |
D3F
|
|
|
🔍 LOW
Missing logs (4)
SIEM rules (3)
Database endpoint connection flood — high volume of new connections
event_type=db_connection_attempt AND dest_port=5432 AND count > 50 within 1m grouped by src_ip
Database connection pool exhaustion — max connections reached
event_type=db_connection_error AND error_code=too_many_connections OR wait_count > 80
Outbound DNS queries from database host to unknown domains
event_type=dns_query AND src_ip=db_host_ip AND dest_domain not in approved_domains
IOCs (3)
|
|||||
4 |
API Gateway | N/A |
AI
|
(AI) API Gateway Request Volume Exhaustion: The API Gateway has redundant controls but lacks rate limiting or request size restrictions. An attacker can send a high volume of requests, exhausting gateway connection pools or thread pools, leading to denial of service for legitimate customers.
TK-E37F16E0
|
0.90
|
|
HIGH (7.6) |
NET
D3F
|
|
|
🔍 MEDIUM
Missing logs (3)
SIEM rules (3)
API Gateway connection pool exhaustion — sustained high request volume
event_type=api_gateway_request AND (response_status=503 OR response_status=504) AND count > 100 within 1m
API Gateway thread pool saturation — high concurrency from single source
event_type=api_gateway_request AND concurrency_metrics > 90th_percentile grouped by src_ip within 5m
API Gateway request size anomalies — abnormally large payloads
event_type=api_gateway_request AND request_body_size > 1048576 grouped by src_ip
IOCs (3)
|
|||||
5 |
secure data zone | N/A |
AI
|
(AI) Lack of Redundancy and DoS Vulnerability: The Secure Data Zone is marked as 'Redundant: No'. As a critical trust boundary with no specified DDoS protections (WAF/IDS/IPS absent), an attacker could flood the segment with traffic, causing unavailability of services relying on this zone. The lack of redundancy means there is no failover capacity.
TK-5773B6BB
|
0.90
|
|
HIGH (7.6) |
NET
D3F
|
|
|
|
🔍 MEDIUM
Missing logs (3)
SIEM rules (3)
Volumetric traffic anomaly to data zone
count(srcIP) > 5000 within 60s AND dstZone=secure_data AND protocol IN (tcp,udp) AND rate > 1000pps
Unusual protocol mix in data zone
entropy(protocol) > 0.8 AND dstZone=secure_data AND duration > 300s
Failed DDoS mitigation bypass
eventName=DDoSAttackDetected AND mitigationStatus=failed AND time > 24h
IOCs (3)
|