Borrowing it
Nothing to install: this file belongs to Lambenthan/empiricalwiki. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/Lambenthan/empiricalwiki/main/.claude/skills/exp-design/SKILL.mdgit clone --depth 1 https://github.com/Lambenthan/empiricalwikiWrote 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/lambenthan/empiricalwiki/exp-design)<a href="https://agentmods.dev/skills/lambenthan/empiricalwiki/exp-design"><img src="https://agentmods.dev/badge/skills/lambenthan/empiricalwiki/exp-design/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/lambenthan/empiricalwiki/exp-design"><img src="https://agentmods.dev/badge/skills/lambenthan/empiricalwiki/exp-design.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 | $0.00045 | $0.04175 |
| Opus 5 | $0.00023 | $0.02087 |
| Sonnet 5 | $0.00009 | $0.00835 |
| Haiku 4.5 | $0.00005 | $0.00417 |
Grade A, and why
exp-design 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 9d 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 — 360 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/exp-design
根据一个 idea(或自由文本假设),设计完整的实验计划。 以 claims 为核心:从 Target / Decomposition / Threats 三个维度界定要验证的 claims, 设计 baseline(基线复现)、validation(核心验证)、ablation(因素隔离)、robustness(鲁棒性)四种实验块。 实验按依赖关系排序,阶段间设决策门(sanity fail → 提前停止)。 可选 Review LLM review 检查实验完整性。所有实验写入 wiki/experiments/ 并添加 graph edges。
Inputs
idea:以下之一:- wiki/ideas/ 中的 slug(如
sparse-lora-for-edge-devices) - 自由文本假设描述(直接提供实验目标)
- wiki/ideas/ 中的 slug(如
--review(可选):启用 Review LLM review 审查实验计划完整性--budget <gpu-hours>(可选):总计算预算上限(GPU 小时),影响 robustness 实验规模
Outputs
wiki/experiments/{slug}.md— 每个实验块一个页面(status: planned)wiki/graph/edges.jsonl— 新增 experiment → claim 的 tested_by 边wiki/ideas/{slug}.md— 更新 linked_experiments 字段wiki/graph/context_brief.md— 重建wiki/graph/open_questions.md— 重建wiki/log.md— 追加日志- EXPERIMENT_PLAN_REPORT(输出到终端)— 实验块总览、执行顺序、计算预算
Wiki Interaction
Reads
wiki/ideas/{slug}.md— 获取 idea 的 hypothesis、approach、risks、origin_gapswiki/claims/*.md— 目标 claims 的当前状态、已有 evidence、confidencewiki/experiments/*.md— 已有实验(避免重复设计、参考 setup 配置)wiki/papers/*.md— 相关论文的 baselines 和实验设置wiki/concepts/*.md— 涉及的技术概念(指导实验设计)wiki/graph/context_brief.md— 全局上下文wiki/graph/open_questions.md— 知识缺口(指导实验优先级)
Writes
wiki/experiments/{slug}.md— 创建实验页面(每个实验块一个)wiki/ideas/{slug}.md— 更新 linked_experiments 字段wiki/graph/edges.jsonl— 添加 tested_by 边wiki/graph/context_brief.md— 重建wiki/graph/open_questions.md— 重建wiki/log.md— 追加操作日志
Graph edges created
tested_by:claim → experiment(claim 被该实验验证)
Workflow
前置:确认工作目录为 wiki 项目根(包含 wiki/、raw/、tools/ 的目录)。
Step 1: 加载上下文
- 解析 idea 输入:
- 若为 slug:读取
wiki/ideas/{slug}.md,提取## Motivation、## Hypothesis、## Approach sketch、## Risks,以及 frontmatter 字段origin_gaps、tags、domain、priority(遵循 CLAUDE.md 的 ideas template) - 若为自由文本:直接作为假设描述使用
- 若为 slug:读取
- 加载相关 wiki 上下文:
- 读取
wiki/graph/context_brief.md(全局上下文) - 读取
wiki/graph/open_questions.md(知识缺口) - 从 idea 的
origin_gaps读取对应的wiki/claims/*.md(目标 claims) - 从每个目标 claim 的
source_papers字段读取对应的wiki/papers/*.md,获取 baseline setup 和已有实验协议 —— 这是 idea → claim → paper 的规范路径(ideas 不带linked_papers字段,改用origin_gaps→source_papers) - 读取已有
wiki/experiments/*.md,检查是否已有类似实验
- 读取
- 若 idea 无 origin_gaps:从假设描述中提取隐含的 claims,在 wiki/claims/ 中查找或标注需要新建
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.
- 9d ago First seen · 360 lines · 45 tokens per session scan A d106f414d4b5
exp-design is a skill published in the GitHub repository Lambenthan/empiricalwiki (82 stars, last pushed 2mo ago), licensed MIT. It adds 45 tokens to every session and 4,175 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
review-paper
Comprehensive manuscript review with three modes: single-pass (default), --adversarial critic-fixer loop, and --peer [journal] simulated peer-review pipeline (editor + 2 dispositioned referees + editorial decision, calibrated to a target journal). R&R continuation via --peer --r2/--r3; hostile-editor stress test via…
audit-reproducibility
Enforce the replication-protocol.md rule by cross-checking numeric claims in a manuscript against the actual R / Stata / Python outputs. Report PASS/FAIL per claim against tolerance thresholds. Use before submission and before releasing a replication package.
capture-environment
Snapshot the computational environment for a replication package — detects the analysis stack (R / Stata / Python) and emits the right lockfiles (renv.lock + sessionInfo.txt, requirements.txt / environment.yml / uv.lock, Stata version + ado package list), records seeds and RNG kind, optionally writes a pinning…
data-management-plan
Draft a funder-compliant Data Management Plan (NSF DMP, NIH DMS Policy 2023, ERC, Horizon Europe) by composing the confidential-data and environment-capture primitives. Sections cover data description, formats/metadata, storage/backup, access/sharing, preservation/archiving, and roles. Use when user says "data…
grant-proposal
Scaffold a research grant proposal (NSF, NIH, ERC, or foundation) by composing existing primitives — pulls identification strategy from an /interview-me spec, delegates the data-management plan to /data-management-plan and the facilities statement to /capture-environment, and emits a funder-requirements checklist. Use…
power-analysis
Compute statistical power, required sample size, and minimum detectable effect (MDE) for a study design, then write a registry-ready power section. Handles two-arm RCTs (with clustering / ICC and unequal allocation), multiple-arm corrections, and a simulation-based power option for non-standard designs…