Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

T1496.004 Cloud Service Hijacking · cyberstrikeus | DecimalAI

cyberstrikeus/T1496.004_cloud-service-hijacking

SkillScore86Maintained

Adversaries may leverage compromised software-as-a-service (SaaS) applications to complete resource-intensive tasks, which may impact hosted service availability.

Securityby cyberstrikeus

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Security & AppSec packmitre-attackenterpriset1496.004

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/cyberstrikeus-t1496-004-cloud-service-hijacking/SKILL.md

decimalai skills pull cyberstrikeus-t1496-004-cloud-service-hijacking --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/cyberstrikeus-t1496-004-cloud-service-hijacking/SKILL.md https://app.decimal.ai/s/cyberstrikeus-t1496-004-cloud-service-hijacking/SKILL.md

Raw file: /s/cyberstrikeus-t1496-004-cloud-service-hijacking/SKILL.md ↗ · machine-readable summary at /s/cyberstrikeus-t1496-004-cloud-service-hijacking.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 23d ago

Yes — +41% lift-10% tokens0% turns86% of whole cases pass with skill (19/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-07	✗→✓	▲ Improved	71%	0%
case-11	✗→✓	▲ Improved	-26%	0%
case-12	✗→✓	▲ Improved	-36%	0%
case-14	✗→✓	▲ Improved	20%	0%
case-15	✗→✓	▲ Improved	-69%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

cyberstrikeus / cyberstrike

.cyberstrike/skill/mitre_attack/TA0040_impact/T1496.004_cloud-service-hijacking/SKILL.md

View on GitHub ↗

License AGPL-3.0Imported 8/10/2026Version v1

Skill contents

SKILL.md

T1496.004 Cloud Service Hijacking

> Sub-technique of: T1496

High-Level Description

Adversaries may leverage compromised software-as-a-service (SaaS) applications to complete resource-intensive tasks, which may impact hosted service availability.

For example, adversaries may leverage email and messaging services, such as AWS Simple Email Service (SES), AWS Simple Notification Service (SNS), SendGrid, and Twilio, in order to send large quantities of spam / Phishing emails and SMS messages. Alternatively, they may engage in LLMJacking by leveraging reverse proxies to hijack the power of cloud-hosted AI models.

In some cases, adversaries may leverage services that the victim is already using. In others, particularly when the service is part of a larger cloud platform, they may first enable the service. Leveraging SaaS applications may cause the victim to incur significant financial costs, use up service quotas, and otherwise impact availability.

Kill Chain Phase

Impact (TA0040)

Platforms: SaaS

What to Check

] Identify if Cloud Service Hijacking technique is applicable to target environment
] Check SaaS systems for indicators of Cloud Service Hijacking
] Assess detection coverage (1 detection strategies)

How to Test

Manual Testing

Identify Attack Surface: Determine if the target environment is susceptible to Cloud Service Hijacking by examining the target platforms (SaaS).

Assess Existing Defenses: Review whether mitigations for T1496.004 are in place. If defenses are absent or misconfigured, this technique may be exploitable.

Execute Test: Use tools and methods described in the MITRE ATT&CK page and external references below.

> Note: No Atomic Red Team tests available for this technique. See Atomic Red Team GitHub for updates.

Remediation Guide

No specific mitigations documented for this technique.

Detection

Detection Strategy for Cloud Service Hijacking via SaaS Abuse

Risk Assessment

| Finding | Severity | Impact | | -------------------------------------------- | -------- | ------ | | Cloud Service Hijacking technique applicable | High | Impact |

CWE Categories

| CWE ID | Title | | ------- | --------------------------------- | | CWE-400 | Uncontrolled Resource Consumption |

References

Illustrative Want to see what cyberstrikeus/T1496.004_cloud-service-hijacking would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/31/2026, 01:19 UTC · 22 cases

Lift

+41%

86% vs 45%

Agent turns

1.0 vs 1.0

Tokens

-10%

2,504.09 vs 2,783.77

fail → pass · 9pass → pass · 10fail → fail · 3

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 We are reviewing our SaaS ecosystem—specifically external messaging APIs like AWS SES and hosted AI endpoints—for unauthorized resource exploitation. Please perform an evaluation of our attack surface and current defenses against SaaS resource hijacking. Provide a report listing the identified exposure points, defense gaps, and overall risk impact rating.	fail→fail	23,457	29,619	+26%	1	1	0%	3,547	4,717	+33%	0	0	—
case-02 I need a security assessment for MITRE ATT&CK technique T1496.004 in our cloud environment. Please inspect our SaaS integrations (including messaging and language model services), detail the steps to manually verify defense posture, and deliver a risk assessment table detailing findings and severity levels.	fail→fail	20,654	33,863	+64%	1	1	0%	3,266	5,504	+69%	0	0	—
case-03 We suspect an attacker compromised AWS IAM keys and launched g4dn.xlarge EC2 instances to run xmrig cryptocurrency miners. Please conduct a security assessment for MITRE ATT&CK technique T1496 Resource Hijacking on IaaS compute instances, including specific MITRE mitigations and Atomic Red Team test procedure IDs.	pass→pass	32,782	30,969	-6%	1	1	0%	3,971	5,136	+29%	0	0	—
case-04 Provide an assessment guide for analyzing compromised user credentials used to log into Salesforce and Microsoft 365 web consoles under MITRE ATT&CK technique T1078.004, including authentication log analysis and MFA enforcement guidance.	pass→pass	34,403	28,750	-16%	1	1	0%	4,681	5,433	+16%	0	0	—
case-05 Analyze an incident where an internal user uploaded 50GB of sensitive source code to a personal Dropbox repository using HTTP POST requests under MITRE ATT&CK technique T1567. Include standard data loss prevention mitigations.	pass→pass	24,965	29,631	+19%	1	1	0%	3,026	4,426	+46%	0	0	—
case-06 When documenting MITRE ATT&CK technique T1496.004 in an application security risk portal, which Common Weakness Enumeration ID applies to unauthorized SaaS resource usage like email blast API abuse? Provide the exact CWE identifier.	pass→pass	22,876	3,344	-85%	1	1	0%	3,028	1,218	-60%	0	0	—
case-07 Our automated adversary emulation framework needs to pull test scripts for T1496.004 (Cloud Service Hijacking). Please provide the command to run the official Atomic Red Team test for this sub-technique.	fail→pass	13,578	13,871	+2%	1	1	0%	1,188	2,035	+71%	0	0	—
case-08 In our MITRE ATT&CK coverage matrix, which enterprise kill chain phase (TA number and name) does sub-technique T1496.004 fall under?	pass→pass	12,395	6,915	-44%	1	1	0%	1,340	1,084	-19%	0	0	—
case-09 Which primary platform target is designated for MITRE ATT&CK technique T1496.004 when mapping infrastructure boundaries?	pass→pass	39,030	9,236	-76%	1	1	0%	6,288	1,073	-83%	0	0	—
case-10 Our compliance team requires a list of official MITRE ATT&CK mitigations for T1496.004 Cloud Service Hijacking to close an audit finding. What specific MITRE mitigation codes (M10xx) should be referenced?	pass→pass	26,536	15,139	-43%	1	1	0%	3,367	2,466	-27%	0	0	—
case-11 We are building a standardized risk assessment finding for SaaS resource hijacking. Provide the exact finding string, severity, and impact classification to record in the risk register when T1496.004 is determined to be applicable.	fail→pass	10,281	2,181	-79%	1	1	0%	1,546	1,151	-26%	0	0	—
case-12 What are the three sequential manual testing phases recommended for evaluating an organization's susceptibility to T1496.004?	fail→pass	18,416	2,889	-84%	1	1	0%	2,038	1,311	-36%	0	0	—
case-13 What key verification items should a cloud security auditor check when evaluating susceptibility to SaaS resource hijacking under T1496.004?	fail→fail	18,307	18,441	+1%	1	1	0%	2,705	3,715	+37%	0	0	—
case-14 An attacker gained unauthorized API keys for hosted Large Language Model services and routed external traffic through reverse proxies to consume model generation tokens. Which specific MITRE ATT&CK sub-technique describes this LLMJacking activity?	fail→pass	11,913	3,040	-74%	1	1	0%	1,113	1,332	+20%	0	0	—
case-15 Threat actors hijacked AWS Simple Email Service (SES) and Twilio accounts to broadcast large-scale phishing campaigns. Under which sub-technique of Resource Hijacking is this SaaS messaging abuse classified?	fail→pass	28,618	4,199	-85%	1	1	0%	4,449	1,377	-69%	0	0	—
case-16 During an investigation, security responders noted that an attacker enabled previously unused cloud messaging features (such as AWS SNS) before sending spam. How does MITRE ATT&CK T1496.004 account for adversary interaction with dormant SaaS services?	pass→pass	26,574	11,987	-55%	1	1	0%	2,803	2,674	-5%	0	0	—
case-17 What business and operational impacts are caused when attackers execute T1496.004 against victim SaaS applications?	pass→pass	21,736	21,497	-1%	1	1	0%	2,531	3,069	+21%	0	0	—
case-18 What is the parent MITRE ATT&CK technique ID and name for T1496.004?	pass→pass	8,942	6,885	-23%	1	1	0%	653	1,146	+75%	0	0	—
case-19 When auditing detection capabilities for SaaS abuse under T1496.004, how many detection strategies are documented in the technique reference, and what is the target focus?	fail→pass	30,454	2,697	-91%	1	1	0%	4,050	1,315	-68%	0	0	—
case-20 Which threat intelligence research documented the LLMJacking threat vector involving reverse proxies and stolen cloud credentials used for AI resource theft?	fail→pass	5,282	6,118	+16%	1	1	0%	770	1,695	+120%	0	0	—
case-21 Which research reports analyzed AWS SES and SNS abuse for messaging spam campaigns in cloud environments?	fail→pass	16,966	12,266	-28%	1	1	0%	2,459	1,861	-24%	0	0	—
case-22 When conducting a T1496.004 gap analysis, what platforms should be examined during step 1 of manual testing to determine attack surface?	fail→pass	33,970	8,674	-74%	1	1	0%	2,424	1,352	-44%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 45.5%
Pass rate with the skill: 86.4%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/31/2026, 01:19 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-07

Prompt

Our automated adversary emulation framework needs to pull test scripts for T1496.004 (Cloud Service Hijacking). Please provide the command to run the official Atomic Red Team test for this sub-technique.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run86

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns-12% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 22 days ago

Other measured skills in the registry, with their headline benchmark lift.