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 Nero1688/claude-academic-skills --skill causal-inference-architectgit clone --depth 1 https://github.com/Nero1688/claude-academic-skillsWrote 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/nero1688/claude-academic-skills/causal-inference-architect)<a href="https://agentmods.dev/skills/nero1688/claude-academic-skills/causal-inference-architect"><img src="https://agentmods.dev/badge/skills/nero1688/claude-academic-skills/causal-inference-architect/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/nero1688/claude-academic-skills/causal-inference-architect"><img src="https://agentmods.dev/badge/skills/nero1688/claude-academic-skills/causal-inference-architect.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 | $0.00456 | $0.01995 |
| Opus 5 | $0.00228 | $0.00997 |
| Sonnet 5 | $0.00091 | $0.00399 |
| Haiku 4.5 | $0.00046 | $0.00199 |
Grade A, and why
causal-inference-architect 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 12d 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 — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
因果推論架構師(Causal Inference Architect)
Step 1|因果問題定式(不清楚就不往下)
寫成一句話:「〔處理 D〕對〔結果 Y〕在〔母體〕的效果,處理時點是〔何時/是否交錯〕, 反事實是〔誰〕」。三個立即檢查:
- D 的變異來源是什麼?(政策強制/自願採用/門檻規則)——變異來源決定策略。
- 自願採用=自選擇,DiD 只處理「時間不變」的選擇,時變選擇要正面交代。
- 預期效果的時間形狀?(立即/漸進/暫時)——決定事件窗與動態設定。
事件日哪裡來(台灣研究):短窗事件研究的事件日,首選公司官方揭露——
MOPS 重大訊息(公司「宣布」那一刻=事件日),乾淨且免費。要取得乾淨事件檔
先走 public-disclosure-scout,它把揭露整成 ticker, event_date, subject 直接餵本 skill。
Step 2|識別策略決策樹
| 變異型態 | 首選策略 | 關鍵假設(寫進論文的話) |
|---|---|---|
| 多單位、不同時點採用 | 交錯 DiD(現代估計量,見 Step 3) | 平行趨勢(條件於共變數) |
| 單一時點政策、全體受影響 | 事前事後+控制組(古典 DiD)/ SCM | 平行趨勢 / 供體池可組合反事實 |
| 連續指標跨門檻觸發處理 | RDD(sharp/fuzzy) | 門檻附近局部隨機、無精準操縱(McCrary 檢定) |
| 有外生工具 | IV/2SLS | 相關性(第一階段 F)+排除限制(只能論證不能檢定) |
| 單一處理單位(一家公司/一個縣市) | 合成控制 SCM | 供體池、前期擬合良好 |
| 都沒有 | 誠實承認做關聯研究,用選擇模型/配對減緩,不宣稱因果 | — |
Step 3|現代 DiD 正典(交錯採用必讀,2020 後革命的核心)
- TWFE 的病:交錯採用+效果異質時,雙向固定效果把「已處理者當控制組」, 權重可為負,估計值可能連正負號都錯。先跑 Goodman-Bacon 分解看壞比較佔多少。
- estimator 選擇(細節與語法見
references/estimator-playbook.md):- Callaway & Sant'Anna(組別×時期 ATT,再聚合)——預設首選,控制組用 never-treated 或 not-yet-treated 要明說。
- Sun & Abraham(交互加權事件研究)——要動態路徑時。
- Stacked DiD——直觀穩健的備援,樣本量夠時。
- 事件研究圖是必需品不是裝飾:前期係數應近 0(pre-trends),圖要含信賴區間; 前期不平行不能只說「大致平行」——用 Rambachan & Roth 誠實區間量化「偏離多少 仍成立」。
- 全部交給 management-figure 出投稿級事件研究圖。
Step 4|診斷與安慰劑矩陣(審稿人一定要的)
| 診斷 | 做法 |
|---|---|
| 平行趨勢 | 事件研究前期係數+聯合檢定;誠實區間敏感度 |
| 安慰劑-時間 | 假處理時點提前 k 年,效果應消失 |
| 安慰劑-單位 | 未處理單位隨機指派假處理(SCM 標配 permutation) |
| 組成穩定 | 進出樣本的公司是否隨處理改變(balanced vs unbalanced 並陳) |
| 預期效應 | 採用「宣布」與「生效」分開檢驗 |
| IV 專屬 | 第一階段 F(弱工具用 effective F);過度識別檢定的侷限誠實說 |
Step 5|審稿人攻防表(輸出必附)
每個識別假設一列:假設|可檢驗?|你的證據|殘餘威脅|回應話術。 「排除限制」這類不可檢驗假設,攻防表要寫制度論證(為什麼工具只透過 D 影響 Y), 不能寫「檢定通過」——沒有這種檢定,亂寫會被識破。
紅線
- 識別不足就不宣稱因果;「關聯+機制佐證」是誠實且可發表的定位。
- 交錯採用場景禁止只報 TWFE;至少一個異質穩健估計量+事件研究圖。
- 每個假設標明「可檢驗/只可論證」;不可檢驗的部分不偽裝成已檢驗。
- 估計量與套件版本寫進再現性聲明(此領域套件更新快,版本影響結果)。
- 用了哪個估計量/套件就正確引用其論文與軟體——清單見
references/method-and-software-citations.md;這既是規範,也是對方法作者最恰當的致謝。
What ships with it
2 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.
- 12d ago First seen · 75 lines · 456 tokens per session scan A d4fc16385a4a
causal-inference-architect is a skill published in the GitHub repository Nero1688/claude-academic-skills (6 stars, last pushed 9d ago), licensed MIT. It adds 456 tokens to every session and 1,995 once invoked, about $0.0023 per session on Opus 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
alterlab-deep-research
Runs a 13-agent deep research pipeline for rigorous academic work on any topic across 7 modes (full research, quick brief, paper review, lit-review, fact-check, Socratic guided research dialogue, and systematic review with optional meta-analysis), covering research-question formulation, Socratic mentoring, methodology…
alterlab-imaging-data-commons
Query and download public cancer imaging data from the NCI Imaging Data Commons (IDC) using the idc-index Python package, filtering by metadata, visualizing in-browser, and checking licenses, with no authentication required. Use when obtaining large-scale radiology (CT, MR, PET) or digital pathology DICOM datasets for…
alterlab-pyhealth
Develops, tests, and deploys clinical machine learning models with the PyHealth healthcare AI toolkit. Use when working with electronic health records (EHR), clinical prediction tasks (mortality, readmission, drug recommendation), medical coding systems (ICD, NDC, ATC), physiological signals (EEG, ECG), healthcare…
alterlab-cobrapy
Build and analyze genome-scale constraint-based metabolic models with COBRApy — flux balance analysis (FBA), flux variability analysis (FVA), gene and reaction knockouts, flux sampling, and SBML model I/O. Use when simulating metabolic networks, predicting growth or knockout phenotypes, or running systems-biology and…
alterlab-deeptools
Process and visualize deep-sequencing coverage with the deepTools CLI — convert BAM to bigWig (bamCoverage), build log2 ratio tracks (bamCompare), run QC (multiBamSummary correlation, PCA, plotFingerprint), apply the ATAC-seq Tn5 shift (alignmentSieve --ATACshift), and make TSS/peak heatmaps and profiles…
alterlab-phylogenetics
Build phylogenetic trees end-to-end from raw sequences — MAFFT multiple sequence alignment, optional TrimAl trimming, IQ-TREE 2 maximum-likelihood inference with model selection and bootstraps, FastTree for large datasets, then visualize with ETE3 or FigTree. Use when reconstructing trees from sequences (FASTA) for…