anomaly-characterization
This Claude Code skill systematically describes and classifies anomalous phenomena that cannot be explained by existing theories, providing precise starting points for abductive reasoning. Use it when you have concrete anomalous observations with a clear reference baseline and need to quantify deviations, eliminate trivial explanations, and categorize the anomaly type before pursuing novel theoretical explanations.
git clone --depth 1 https://github.com/yogsoth-ai/de-anthropocentric-research-engine /tmp/anomaly-characterization && cp -r /tmp/anomaly-characterization/skills/anomaly-characterization ~/.claude/skills/anomaly-characterizationSKILL.md
# Anomaly Characterization
系统描述和分类异常现象，为溯因推理（abductive reasoning）提供精确的起点。
## HARD-GATE
<HARD-GATE>
前置条件（全部满足才能开始）:
1. 已有具体的异常观察描述（不能是模糊的"结果很奇怪"）
2. 有参照基准（预期结果或理论预测）用于量化偏差
不满足 → 停止，返回错误：异常描述不足，需要具体观察和参照基准。
</HARD-GATE>
## Pipeline
1. 前置检查：验证异常描述和参照基准完整性
2. 现象描述：用精确语言重述异常（what was observed vs. what was expected）
3. 与预期偏差量化：量化或定性描述偏差程度（magnitude, direction, frequency）
4. 排除已知解释：列举并逐一排除可能的平凡解释（测量误差、采样偏差、已知效应）
5. 异常分类：将异常归类（unexpected absence / unexpected presence / unexpected magnitude / unexpected pattern / unexpected timing）
6. 输出结构化异常描述
## Output Format
```json
{
"anomaly_id": "A1",
"phenomenon": "Precise description of what was observed",
"expected": "What theory or prior evidence predicted",
"deviation": {
"direction": "higher | lower | absent | present | different_pattern",
"magnitude": "Quantitative or qualitative estimate",
"frequency": "Isolated | recurring | systematic"
},
"excluded_explanations": [
{"explanation": "...", "reason_excluded": "..."}
],
"anomaly_type": "unexpected_absence | unexpected_presence | unexpected_magnitude | unexpected_pattern | unexpected_timing",
"severity": "minor | moderate | major",
"notes": "Additional context"
}
```Experiment-specific - summarize the DARE executor's research design into a clean research_result report, forced to write back into the spec file produced by formated-specs.
Experiment-specific - replaces writing-specs, emits DARE's 4-layer call plan as a clean research_graph schema. Last step forces load formated-result.
loss-1 judge - read a sample's full dialogue and decide whether the user simulator semantically enacted its Policy Card. check-blind.
loss-2 judge - pairwise quality comparison across the n rungs within one topic; decide monotonicity and endpoint separation. check-blind, D1-D5 only.
Strategy: 面对异常的最佳解释推理
Remove components one by one, observe system changes to reveal hidden dependencies and generate ideas from structural gaps.
Map system architecture to ablatable units for ablation studies
Design ablation studies to isolate component contributions in ML systems