Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Adversaries may launch a denial of service (DoS) attack targeting an endpoint's operating system (OS).
.claude/skills/cyberstrikeus-t1499-001-os-exhaustion-flood/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -32% | 0% |
> Sub-technique of: T1499
Adversaries may launch a denial of service (DoS) attack targeting an endpoint's operating system (OS). A system's OS is responsible for managing the finite resources as well as preventing the entire system from being overwhelmed by excessive demands on its capacity. These attacks do not need to exhaust the actual resources on a system; the attacks may simply exhaust the limits and available resources that an OS self-imposes.
Different ways to achieve this exist, including TCP state-exhaustion attacks such as SYN floods and ACK floods. With SYN floods, excessive amounts of SYN packets are sent, but the 3-way TCP handshake is never completed. Because each OS has a maximum number of concurrent TCP connections that it will allow, this can quickly exhaust the ability of the system to receive new requests for TCP connections, thus preventing access to any TCP service provided by the server.
ACK floods leverage the stateful nature of the TCP protocol. A flood of ACK packets are sent to the target. This forces the OS to search its state table for a related TCP connection that has already been established. Because the ACK packets are for connections that do not exist, the OS will have to search the entire state table to confirm that no match exists. When it is necessary to do this for a large flood of packets, the computational requirements can cause the server to become sluggish and/or unresponsive, due to the work it must do to eliminate the rogue ACK packets. This greatly reduces the resources available for providing the targeted service.
Platforms: Linux, macOS, Windows
> Note: No Atomic Red Team tests available for this technique. See Atomic Red Team GitHub for updates.
Leverage services provided by Content Delivery Networks (CDN) or providers specializing in DoS mitigations to filter traffic upstream from services. Filter boundary traffic by blocking source addresses sourcing the attack, blocking ports that are being targeted, or blocking protocols being used for transport. To defend against SYN floods, enable SYN Cookies.
| Finding | Severity | Impact | | ---------------------------------------- | -------- | ------ | | OS Exhaustion Flood technique applicable | Low | Impact |
| CWE ID | Title | | ------- | --------------------------------- | | CWE-400 | Uncontrolled Resource Consumption |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 33,521 | 41,145 | +23% | 1 | 1 | 0% | 4,452 | 7,037 | +58% | 0 | 0 | — |
case-02 | fail→fail | 45,446 | 39,253 | -14% | 1 | 1 | 0% | 6,611 | 5,878 | -11% | 0 | 0 | — |
case-03 | pass→pass | 34,942 | 28,346 | -19% | 1 | 1 | 0% | 4,015 | 5,117 | +27% | 0 | 0 | — |
case-04 | pass→pass | 27,356 | 23,542 | -14% | 1 | 1 | 0% | 3,020 | 4,357 | +44% | 0 | 0 | — |
case-05 | pass→pass | 29,839 | 23,122 | -23% | 1 | 1 | 0% | 3,218 | 4,034 | +25% | 0 | 0 | — |
case-06 | pass→fail | 29,961 | 23,415 | -22% | 1 | 1 | 0% | 3,218 | 4,073 | +27% | 0 | 0 | — |
case-07 | fail→fail | 27,050 | 26,475 | -2% | 1 | 1 | 0% | 2,867 | 3,887 | +36% | 0 | 0 | — |
case-08 | pass→pass | 5,400 | 4,937 | -9% | 1 | 1 | 0% | 1,005 | 1,613 | +60% | 0 | 0 | — |
case-09 | pass→pass | 9,069 | 4,878 | -46% | 1 | 1 | 0% | 1,113 | 1,623 | +46% | 0 | 0 | — |
case-10 | fail→pass | 7,068 | 10,368 | +47% | 1 | 1 | 0% | 1,207 | 1,870 | +55% | 0 | 0 | — |
case-11 | fail→pass | 7,098 | 4,635 | -35% | 1 | 1 | 0% | 1,164 | 1,799 | +55% | 0 | 0 | — |
case-12 | pass→pass | 9,347 | 5,462 | -42% | 1 | 1 | 0% | 1,477 | 1,855 | +26% | 0 | 0 | — |
case-13 | pass→pass | 20,748 | 16,360 | -21% | 1 | 1 | 0% | 3,416 | 3,520 | +3% | 0 | 0 | — |
case-14 | fail→pass | 21,409 | 4,040 | -81% | 1 | 1 | 0% | 2,584 | 1,507 | -42% | 0 | 0 | — |
case-15 | pass→pass | 15,416 | 17,756 | +15% | 1 | 1 | 0% | 2,339 | 3,999 | +71% | 0 | 0 | — |
case-16 | pass→pass | 24,457 | 23,824 | -3% | 1 | 1 | 0% | 2,920 | 3,540 | +21% | 0 | 0 | — |
case-17 | pass→pass | 19,200 | 13,572 | -29% | 1 | 1 | 0% | 2,223 | 3,079 | +39% | 0 | 0 | — |
case-18 | pass→pass | 7,931 | 8,899 | +12% | 1 | 1 | 0% | 1,224 | 2,218 | +81% | 0 | 0 | — |
case-19 | fail→fail | 20,158 | 25,013 | +24% | 1 | 1 | 0% | 2,923 | 4,164 | +42% | 0 | 0 | — |
case-20 | pass→pass | 9,838 | 7,451 | -24% | 1 | 1 | 0% | 848 | 1,523 | +80% | 0 | 0 | — |
case-21 | fail→pass | 21,107 | 4,056 | -81% | 1 | 1 | 0% | 2,075 | 1,406 | -32% | 0 | 0 | — |
case-22 | pass→pass | 7,392 | 7,340 | -1% | 1 | 1 | 0% | 1,171 | 1,468 | +25% | 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. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.