Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Adversaries may use steganography techniques in order to prevent the detection of hidden information.
.claude/skills/cyberstrikeus-t1406-001-steganography/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -62% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 17% | 0% |
> Sub-technique of: T1406
Adversaries may use steganography techniques in order to prevent the detection of hidden information. Steganographic techniques can be used to hide data in digital media such as images, audio tracks, video clips, or text files.
Platforms: Android
Determine if the target mobile environment is susceptible to Steganography by examining the target platforms (Android).
Review whether mitigations for T1406.001 are in place. If defenses are absent or misconfigured, this technique may be exploitable.
No specific mitigations documented for this technique.
| Finding | Severity | Impact | | ---------------------------------- | -------- | --------------- | | Steganography technique applicable | Low | Defense Evasion |
| CWE ID | Title | | ------- | ---------------------------- | | CWE-693 | Protection Mechanism Failure |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 27,809 | 25,713 | -8% | 1 | 1 | 0% | 3,001 | 3,537 | +18% | 0 | 0 | — |
case-02 | fail→fail | 63,890 | 32,772 | -49% | 1 | 1 | 0% | 4,397 | 4,776 | +9% | 0 | 0 | — |
case-03 | fail→pass | 20,656 | 28,276 | +37% | 1 | 1 | 0% | 3,228 | 3,884 | +20% | 0 | 0 | — |
case-04 | pass→pass | 26,081 | 23,921 | -8% | 1 | 1 | 0% | 3,481 | 3,710 | +7% | 0 | 0 | — |
case-05 | pass→pass | 29,311 | 32,604 | +11% | 1 | 1 | 0% | 4,460 | 4,699 | +5% | 0 | 0 | — |
case-06 | pass→pass | 20,734 | 28,532 | +38% | 1 | 1 | 0% | 3,135 | 3,924 | +25% | 0 | 0 | — |
case-07 | pass→pass | 20,238 | 19,171 | -5% | 1 | 1 | 0% | 2,278 | 2,573 | +13% | 0 | 0 | — |
case-08 | fail→pass | 25,592 | 9,722 | -62% | 1 | 1 | 0% | 2,821 | 960 | -66% | 0 | 0 | — |
case-09 | fail→pass | 18,573 | 10,769 | -42% | 1 | 1 | 0% | 2,828 | 1,943 | -31% | 0 | 0 | — |
case-10 | fail→pass | 14,814 | 2,771 | -81% | 1 | 1 | 0% | 1,764 | 663 | -62% | 0 | 0 | — |
case-11 | pass→pass | 9,040 | 7,829 | -13% | 1 | 1 | 0% | 467 | 632 | +35% | 0 | 0 | — |
case-12 | pass→pass | 15,059 | 1,809 | -88% | 1 | 1 | 0% | 1,533 | 621 | -59% | 0 | 0 | — |
case-13 | fail→pass | 25,376 | 27,637 | +9% | 1 | 1 | 0% | 3,258 | 3,804 | +17% | 0 | 0 | — |
case-14 | pass→pass | 23,897 | 28,557 | +20% | 1 | 1 | 0% | 3,549 | 3,442 | -3% | 0 | 0 | — |
case-15 | pass→pass | 16,483 | 8,495 | -48% | 1 | 1 | 0% | 1,728 | 1,068 | -38% | 0 | 0 | — |
case-16 | pass→pass | 12,573 | 2,044 | -84% | 1 | 1 | 0% | 1,397 | 670 | -52% | 0 | 0 | — |
case-17 | pass→pass | 15,902 | 6,566 | -59% | 1 | 1 | 0% | 1,616 | 574 | -64% | 0 | 0 | — |
case-18 | fail→pass | 32,300 | 7,903 | -76% | 1 | 1 | 0% | 3,619 | 633 | -83% | 0 | 0 | — |
case-19 | fail→pass | 20,020 | 8,880 | -56% | 1 | 1 | 0% | 1,814 | 643 | -65% | 0 | 0 | — |
case-20 | fail→pass | 22,798 | 6,878 | -70% | 1 | 1 | 0% | 3,387 | 626 | -82% | 0 | 0 | — |
case-21 | fail→pass | 11,429 | 9,568 | -16% | 1 | 1 | 0% | 1,005 | 761 | -24% | 0 | 0 | — |
case-22 | fail→pass | 23,147 | 18,079 | -22% | 1 | 1 | 0% | 2,337 | 2,360 | +1% | 0 | 0 | — |
case-23 | fail→pass | 16,359 | 32,896 | +101% | 1 | 1 | 0% | 1,307 | 810 | -38% | 0 | 0 | — |
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. 23 cases were attempted. The headline lift of +48 percentage points is the difference between those two pass rates over the 23 comparable cases.
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.
Other measured skills in the registry, with their headline benchmark lift.