Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add tetckx/deep-structural-analysis-skill --skill deep-structural-analysisgit clone --depth 1 https://github.com/tetckx/deep-structural-analysis-skillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tetckx/deep-structural-analysis-skill/deep-structural-analysis)<a href="https://agentmods.dev/skills/tetckx/deep-structural-analysis-skill/deep-structural-analysis"><img src="https://agentmods.dev/badge/skills/tetckx/deep-structural-analysis-skill/deep-structural-analysis/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tetckx/deep-structural-analysis-skill/deep-structural-analysis"><img src="https://agentmods.dev/badge/skills/tetckx/deep-structural-analysis-skill/deep-structural-analysis.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00167 | $0.12901 |
| Opus 5.5 | $0.00067 | $0.05160 |
| Sonnet 5.5 | $0.00033 | $0.02580 |
| Haiku 4.5 | $0.00017 | $0.01290 |
Grade A, and why
deep-structural-analysis scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 378 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deep Structural Analysis
When to Use This Skill
Single-tool application: 单工具请求（"用三向分析一下"）时，输出仅该工具应用（简短），后附"需要我以此工具为入口展开完整的多透镜分析吗？"——不启动完整框架。
致害提醒：以下场景中，即使分析在学术上高质量也可能构成伤害——这不是"分析质量"问题，是"分析行为本身"在权力不对称中的效应问题。Skill 无内置检测机制，依赖使用者判断——遇到此类场景优先考虑"不分析"：
- 未经同意的创伤分析：将未同意者的创伤经历作为"分析案例"——伤害=将受害者转化为展示制度失效的案例
- 权力上位者的武器化使用：制度权力者分析"为什么被统治者不服从"——伤害=分析提供更精密的控制工具
- 紧急情境的分析延迟：安全/医疗危机需立即行动——伤害=分析的结构完整性=决策延迟
- 心理健康危机的误用：急性心理危机以"深度分析"形式表达——伤害=需要连接而非分析
- 活人被非人化：分析具体可识别的个体——伤害=分析行为改变被分析者的社会处境
创伤敏感标准（话题涉系统性暴力/歧视/具体人身伤害时触发，五约束）：
- 前置声明：分析开头加一句"这不是抽象谜题——有人在这件事里真实地受过伤害"，一句足够
- 禁止虚假平衡：不把施害者的"感到被攻击"和受害者的"实际伤害"同等权重呈现——虚假平衡在不对称权力环境中本身就是伤害
- 个体策略约束：不在个体策略行要求受害者"理解对方"或"换位思考"——"互相理解"在安全不对称下不是中性美德
- 尊严优先于完整：无法在不造成二次伤害的情况下完成分析时，优先保护尊严而非分析完整性，显式声明省略了什么——可省略的判定：细节还原/暴力场景描述/受害动机归因解释，且对核心判断非必要
- 反淡化约束：本标准增加精确性而非稀释锐度——一句前置声明已足够，不反复道歉。"指出这个系统让某一方受益于另一方的痛苦"不是二次伤害，是分析存在的理由
过度抽象化检查（同上触发）：自问"这个分析是否在结构性拆解中，把一个植根于真实痛苦的话题变成无血无肉的智力谜题？"——若是：加至少一句不被任何透镜中介的直接承认（"这是真的，这发生在真实的人身上，痛苦不是变量"）；若结构语言使受害方经验不可见，重写相关段落。
情绪中立 = 偏向：权力不对称话题上，"保持绝对情绪中立/不偏不倚"本身就是偏向——它偏向现状，而现状已偏向权力上位者。对称化呈现不对称权力下的经验，是已经在做一个有偏向的操作。这不是"偏袒一方"——是纠正一个已经偏袒的默认状态。
触发边界：结构性不平等但不涉具体个体创伤（收入差距/教育资源分配）→ 不触发；历史事件无直接受害者（已废止制度的回顾）→ 不触发；用户明确"不要顾及感受直接分析"→ 用户自主退出保护层。
Trigger Guard: Complexity Quick Assessment
正向（需要本框架）：≥2 个利益冲突方 / 系统性矛盾（政策 vs 实践）/ 跨多领域 / 用户显式要求结构或多角度分析。 负向（不需要）：事实性解释（"为什么天空是蓝色的"）/ 调试 / 总结 / 信息查询 / 单信源可完全回答。
Fallback：触发词在但复杂度评估为负 → 先给简洁事实回答，再提供展开选项。 Ambiguity：边界情况问一个问题："This could be a short answer or a deeper structural analysis — which would you prefer?" Refusal：两次拒绝澄清后默认低投入选项。协议冲突时：创伤敏感 > 用户显式指令 > 其他默认行为。
Exit / Degradation Protocol
Collapsed Mode（用户说"简单点"/"说人话"/"简短回答"时立即切换）：①核心发现（2-3 句）②⚠️ 关键盲点（若存在会实质改变个体策略的——最多一句；创伤敏感话题严禁对称句式）③分层影响表 ④一个追问"需要我展开某一透镜或工具的详细分析吗？"。Ultra-Collapsed（"再简单点"）：一句话核心发现 + 一条可行动建议（≤20 字）+ 一条风险提示（如适用），纯文本无表格。
Refusal 处理（两次澄清后用户仍"你自己判断"）：默认低投入选项——短回答（≤100 中文字符纯段落，无板块标题）用于个人/临床维度（"我为什么总是失眠"）；聚焦深度（3-4 透镜+执行摘要）用于个人问题触及结构域（"我为什么存不下钱"——嵌入工资/通胀/消费文化）；模糊时默认聚焦。声明"我将采用[简短回答/聚焦深度分析]，如需更深入请随时告知"。用户后续说"展开"直接升级深度，不再问澄清问题。
What ships with it
8 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 378 lines · 167 tokens per session scan A 565fb67ed1e5
deep-structural-analysis is a skill published in the GitHub repository tetckx/deep-structural-analysis-skill (6 stars, last pushed 1mo ago), licensed MIT. It adds 167 tokens to every session and 12,901 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
ultimate-seo-geo
Audits and optimizes websites for search engine visibility (SEO) and AI search citation (GEO), covering technical health, E-E-A-T content scoring, domain authority, structured data, rich results, and entity signals. Use when running SEO audits, diagnosing traffic drops or ranking losses, generating Schema.org JSON-LD…
godfly
Adversarial collaborator who pushes back like a senior engineer - the default adversarial reviewer. Surfaces hidden assumptions, grounds every challenge in evidence, presents real counterpoints with proven alternatives, drives toward synthesis, and records settled claims as verdict nodes in docs/verdicts/ so an…
morpheus
Second opinion and final verdict for engineering decisions. Challenges a plan, design, PR, claim, fix, or bug theory with cited evidence, steelmans the user's position first, asks when an assumption would flip the verdict, labels how each claim is known, and always lands on a verdict or a recommended option with…
gauntlet
Empirically test whether a skill actually improves model output — before trusting it. Runs a controlled experiment: planted-flaw fixture, no-skill control arm, skill arm(s), optional cross-model arms via installed CLIs, blind judging with shuffled labels, and a pressure test for verdict stability. Produces…
mean-qa
Adversarial QA campaigns with persistent .proof evidence and re-runnable cases. Use to QA a feature, test a browser flow, verify a change or staging build, design a test campaign, or prepare a PR verification section. Local evidence is mandatory; publishing is a separate authorized action.
competing-hypotheses
Map the full solution space and evaluate approaches against evidence. Based on the CIA's Analysis of Competing Hypotheses (ACH). Use for DECISIONS - choosing between technologies, architectures, or strategies - or when the user needs counterpoints and alternatives. Triggers on "what are the alternatives", "compare…