Install any skill in seconds. Free to start, no credit card required.
Get Started Free →蓝队与紫队工程:检测规则编写、SIEM/EDR 调优、事件响应、数字取证、威胁狩猎、ATT&CK 映射、紫队演练闭环。Use when writing Sigma/YARA detection rules, tuning SIEM noise, responding to security incidents, conducting forensic analysis, hunting threats, or running purple team exercises.
.claude/skills/telagod-detecting-and-responding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 62% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -20% | 0% |
> 判断先于执行:决定「是否做 / 选什么 / 如何取舍」(栈、方案、架构、权衡)前,先读领域判断内核 skills/_kernel/security/SKILL.md——它管 judgment,本秘典管 execution;冲突时以内核判断为准。
> 检测是工程,不是运气。每条规则必须能回答四问:what / why / FP rate / response。 > 站在防御侧,把告警当代码维护、把事件当事故管理、把狩猎当假设验证。 > 信级:项目日志/EDR 原始事件 > Sigma/YARA 规则库 > ATT&CK 官方矩阵 > 训练记忆(标 [unverified])。
| 意图 | 秘典 | 触发词 | |------|------|--------| | SIEM/EDR 规则与调优 | siem-and-edr | Sigma, YARA, Splunk, Elastic, Sentinel, EDR, LOLBins, detection-as-code | | 事件响应与取证 | incident-response | IR, NIST 800-61, triage, chain of custody, Volatility, memory, runbook, postmortem | | 威胁狩猎与紫队 | threat-hunting | hunt, hypothesis, IOC, IOA, TTP, ATT&CK, Atomic Red Team, Caldera, 蜜罐 |
检测:日志源 → 规则编写 → 告警分级 → 调优降噪 → 覆盖矩阵
响应:识别 → 遏制 → 根因 → 清除 → 恢复 → 复盘
狩猎:假设 → 数据源 → 验证 → 规则化 → 自动化 → 紫队闭环每环必须可回答「我看的是哪条日志?我证伪的是哪条假设?我下一步动作是什么?」
| 场景 | 用 | 不用 | |------|----|----| | 写 Sigma/YARA 规则、调 SIEM | ✅ siem-and-edr | — | | 处理已发生入侵、取证 | ✅ incident-response | — | | 假设驱动狩猎 / 紫队演练 | ✅ threat-hunting | — | | ATT&CK 检测覆盖打分 | ✅ threat-hunting | — | | 设计应用层防御代码 | ❌ | defending-applications | | 渗透测试、写 PoC | ❌ | securing-systems (pentest/red-team) | | 威胁建模、IAM 架构 | ❌ | architecting-security | | 代码静态扫描胶水 | ❌ | analyzing-security | | 云配置基线、K8s 加固 | ❌ | securing-cloud-and-supply-chain |
192.0.2.0/24、域名用 example.com、用户名用 <analyst>。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 24,488 | 25,534 | +4% | 1 | 1 | 0% | 3,447 | 5,556 | +61% | 0 | 0 | — |
case-02 | fail→pass | 26,912 | 22,917 | -15% | 1 | 1 | 0% | 4,656 | 5,054 | +9% | 0 | 0 | — |
case-03 | fail→fail | 27,662 | 31,623 | +14% | 1 | 1 | 0% | 4,741 | 5,723 | +21% | 0 | 0 | — |
case-04 | pass→pass | 19,423 | 24,543 | +26% | 1 | 1 | 0% | 3,505 | 5,488 | +57% | 0 | 0 | — |
case-05 | pass→pass | 34,261 | 19,849 | -42% | 1 | 1 | 0% | 3,500 | 4,919 | +41% | 0 | 0 | — |
case-06 | pass→fail | 23,747 | 33,451 | +41% | 1 | 1 | 0% | 3,748 | 6,686 | +78% | 0 | 0 | — |
case-07 | pass→pass | 17,844 | 20,224 | +13% | 1 | 1 | 0% | 2,726 | 4,440 | +63% | 0 | 0 | — |
case-08 | pass→pass | 19,271 | 19,795 | +3% | 1 | 1 | 0% | 2,971 | 4,049 | +36% | 0 | 0 | — |
case-09 | fail→pass | 16,247 | 16,188 | -0% | 1 | 1 | 0% | 2,400 | 3,480 | +45% | 0 | 0 | — |
case-10 | fail→fail | 18,215 | 23,438 | +29% | 1 | 1 | 0% | 2,620 | 4,967 | +90% | 0 | 0 | — |
case-11 | pass→pass | 20,185 | 25,700 | +27% | 1 | 1 | 0% | 2,790 | 4,704 | +69% | 0 | 0 | — |
case-12 | pass→pass | 18,803 | 25,783 | +37% | 1 | 1 | 0% | 2,914 | 5,638 | +93% | 0 | 0 | — |
case-13 | fail→pass | 20,640 | 27,819 | +35% | 1 | 1 | 0% | 2,991 | 4,858 | +62% | 0 | 0 | — |
case-14 | fail→fail | 26,816 | 25,087 | -6% | 1 | 1 | 0% | 3,637 | 5,256 | +45% | 0 | 0 | — |
case-15 | fail→pass | 26,190 | 30,059 | +15% | 1 | 1 | 0% | 4,137 | 6,221 | +50% | 0 | 0 | — |
case-16 | pass→pass | 24,529 | 29,234 | +19% | 1 | 1 | 0% | 3,324 | 5,494 | +65% | 0 | 0 | — |
case-17 | fail→fail | 20,409 | 16,193 | -21% | 1 | 1 | 0% | 3,632 | 3,574 | -2% | 0 | 0 | — |
case-18 | fail→pass | 13,763 | 3,927 | -71% | 1 | 1 | 0% | 2,315 | 1,850 | -20% | 0 | 0 | — |
case-19 | pass→pass | 20,058 | 25,687 | +28% | 1 | 1 | 0% | 2,745 | 5,071 | +85% | 0 | 0 | — |
case-20 | pass→pass | 12,382 | 5,260 | -58% | 1 | 1 | 0% | 1,765 | 2,003 | +13% | 0 | 0 | — |
case-21 | pass→pass | 13,979 | 12,634 | -10% | 1 | 1 | 0% | 2,137 | 2,897 | +36% | 0 | 0 | — |
case-22 | pass→pass | 15,276 | 8,664 | -43% | 1 | 1 | 0% | 2,296 | 2,502 | +9% | 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.