Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/csmar432/finai-researchnpx agentmods add skills/csmar432/finai-research/fin-idea-discoveryWrote 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/csmar432/finai-research/fin-idea-discovery)<a href="https://agentmods.dev/skills/csmar432/finai-research/fin-idea-discovery"><img src="https://agentmods.dev/badge/skills/csmar432/finai-research/fin-idea-discovery/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/csmar432/finai-research/fin-idea-discovery"><img src="https://agentmods.dev/badge/skills/csmar432/finai-research/fin-idea-discovery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00055 | $0.05162 |
| Opus 5 | $0.00028 | $0.02581 |
| Sonnet 5 | $0.00011 | $0.01032 |
| Haiku 4.5 | $0.00006 | $0.00516 |
Grade A, and why
fin-idea-discovery 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 — 568 lines — stays where its author put it; the contents beside it link to each section on GitHub.
经济金融研究想法发现流程
从研究方向 $ARGUMENTS 开始，经过系统化流程，输出经过数据验证的可执行研究方案。
流程概览
研究方向输入
↓
阶段1: 研究方向理解 — 解析研究领域（绿色金融/数字金融/ESG/碳经济学/宏观金融/公司金融）
↓
阶段2: 文献综述 — 使用 MCP 工具搜索 OpenAlex/ArXiv/NBER/中文顶刊
↓
阶段3: 研究缺口识别 — 从文献中识别 3-5 个具体研究缺口
↓
阶段4: 想法生成 — 从缺口生成 8-12 个研究想法
↓
阶段5: 新颖性预检查 — 对每个想法快速检索 arXiv/NBER
↓
阶段6: 【强制】想法-数据交叉验证 — 使用 idea_data_checker.py 验证每个想法的数据可行性
↓ checkpoint（必须暂停，展示数据可行性表格给用户）
↓
阶段7: 排序输出 — 按新颖性(40%) + 数据可行性(30%) + 发表潜力(30%) 排序
↓
输出: IDEA_REPORT.md + IDEA_DATA_CHECK.md
核心原则
数据优先原则（必须遵守）
数据验证必须前移到想法阶段，不等到数据获取阶段才发现无数据
传统流程（有问题）:
想法生成 → 新颖性验证 → 实证设计 → 数据获取 ← 到这里才发现无数据！
↓ ↓
浪费大量时间 不得不返回更换主题
改进流程（当前）:
想法生成 → 【想法-数据交叉验证】→ 新颖性验证 → 实证设计 → 数据获取
↓ ↓
在此处检查数据可行性 数据已知可行，只需执行
无数据→立即告知用户 预先设计的获取方案
强制 checkpoint
阶段6（数据验证）完成后，必须暂停并展示数据可行性表格给用户，在用户确认前不得进入下一阶段。
输出文件
output/fin-ideas/
├── IDEA_REPORT.md ← 完整想法报告（包含所有想法的详细信息）
├── IDEA_DATA_CHECK.md ← 数据可行性报告（阶段6输出）
└── IDEA_CANDIDATES.md ← 精简版（TOP 3-5 最优想法）
output/fin-novelty/
└── NOVELTY_PRECHECK.md ← 初步新颖性检查结果
阶段详解
阶段1: 研究方向理解
1.1 解析研究领域
根据用户描述，识别研究方向所属领域：
| 领域 | 核心关键词 | 典型数据需求 |
|---|---|---|
| 绿色金融 | ESG、碳排放、绿色债券、气候风险 | ESG评级、碳排放数据、财务面板 |
| 数字金融 | Fintech、数字普惠、移动支付、互联网金融 | 第三方支付数据、用户规模 |
| 碳经济学 | 碳交易、碳配额、碳关税、减排 | 碳市场数据、企业排放数据 |
| 宏观金融 | 货币政策、金融周期、系统性风险 | 宏观指标、金融市场数据 |
| 公司金融 | 融资约束、资本结构、并购、股利政策 | 财务面板、公司治理数据 |
| 资产定价 | 因子模型、异常收益、机构投资者 | 市场数据、因子数据 |
| 行为金融 | 投资者情绪、散户行为、羊群效应 | 交易数据、账户数据 |
| 金融科技 | 区块链、数字货币、API金融 | 平台数据、交易数据 |
1.2 提取用户约束
从用户输入中提取：
- 目标期刊：JF/JFE/RFS/经济研究/金融研究等
- 研究类型：实证/理论/综述/方法创新
- 偏好方法：DID/IV/RDD/机器学习等
- 数据偏好：A股/美股/全球/宏观
- 时间范围：样本期要求
阶段2: 文献综述
2.1 MCP 多源检索
必须按顺序执行以下检索：
# 第1步：NBER 工作论文（预印本先行）
CallMcpTool: user-nber-wp -> search_nber_papers
query: "[核心关键词] + A股/China + 实证方法"
year_from: 2023
# 第2步：OpenAlex 学术论文
CallMcpTool: user-openalex -> get_openalex_works
query: "[研究领域] + [核心机制] + China"
per_page: 30
# 第3步：中文顶刊（A股研究必查）
CallMcpTool: user-brave-search -> brave_web_search
query: "经济研究 金融研究 管理世界 [核心关键词] A股"
num_results: 10
CallMcpTool: user-brave-search -> brave_web_search
query: "中国工业经济 世界经济 [核心机制] 实证"
num_results: 10
# 第4步：ArXiv 预印本（机器学习/计量方法）
CallMcpTool: user-arxiv -> semantic_search
query: "[研究领域] + China + empirical"
max_results: 20
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 · 568 lines · 55 tokens per session scan A 0f0752a85ffe
fin-idea-discovery is a skill published in the GitHub repository csmar432/finai-research (100 stars, last pushed 3d ago), licensed MIT. It adds 55 tokens to every session and 5,162 once invoked, about $0.0003 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
r-econometrics
Run IV, DiD, and RDD analyses in R with proper diagnostics.
econometrics-phd-level
A guide to econometrics, the use of statistics to study relationships in data, based on a 12-part Korean lecture series. It routes questions to explanations of topics such as regression, panel data, instrumental variables, and causal comparisons.
did-causal
Use this Skill when the user needs to estimate causal treatment effects using difference-in-differences (DID) designs: two-way fixed effects (TWFE) regression, parallel trends pre-testing, Callaway-Sant'Anna staggered adoption estimator, and Goodman-Bacon decomposition. Covers both Python (linearmodels) and R (did…
r-econometrics
Generates rigorous, modern, reproducible R code for causal inference and panel econometrics with fixest, heterogeneity-robust DiD estimators (Callaway-Sant'Anna, Sun-Abraham, BJS, de Chaisemartin-D'Haultfoeuille), weak-IV-robust inference, optimal-bandwidth RDD via rdrobust, and wild cluster bootstrap. Use when the…
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.
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…