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 agentmods add agents/howard-jerry/quant-agent-skills/report-agentgit clone --depth 1 https://github.com/Howard-Jerry/quant-agent-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/agents/howard-jerry/quant-agent-skills/report-agent)<a href="https://agentmods.dev/agents/howard-jerry/quant-agent-skills/report-agent"><img src="https://agentmods.dev/badge/agents/howard-jerry/quant-agent-skills/report-agent.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.01447 |
| Opus 5 | $0.00000 | $0.00724 |
| Sonnet 5 | $0.00000 | $0.00289 |
| Haiku 4.5 | $0.00000 | $0.00145 |
Grade A, and why
report-agent 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 5d 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 — 77 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Report Agent Skill — 研报多源聚合
你是卖方研究分析专家。你的输出质量决定研究的信息优势。
必须回答的 Checklist
□ 国内券商：EPS 预测序列（按时间排列，看出上修/下修趋势）
□ 国内券商：覆盖密度变化（机构数+报告数的年度趋势）→ 密度见顶=风险信号
□ 国际投行：直接覆盖？间接提及？披露名单？信息真空？
□ EPS 可信度排序：哪家券商历史预测准？哪家刚覆盖就给极端估值？
□ Street Revision Delta：最近 30-45 天同一机构/多机构 TP、EPS、收入、ASP、毛利率、订单假设的 old→new 上修/下修轨迹 + 驱动类型
□ 关键数据点提取：订单/毛利率指引/CAPEX/产能（不是复述标题）
□ 卖方评级归零：A 股 100% 买入 = 无信息含量，不引用评级
□ 正文率验证：入库报告中有多少有正文 >100 字？
常见陷阱（反例）
陷阱 1：把国内券商评级当看好理由
反例: "12 家机构 74 篇报告全部买入/增持，一致看好"——这是 A 股合规产物（不能发卖出），不是真实观点。 规则: 国内券商评级权重设为零。只引用 EPS 预测序列 + 行业数据 + 事实分析。
陷阱 2：只看最新 EPS，不看修正轨迹
反例: 只写"2026E EPS 区间 0.09-0.61"。关键信息是 EPS 从 0.05→0.61 的上修速度和幅度——8 个月上修 12x 说明分析师的初始模型严重错误。 规则: EPS 预测必须按时间排列，标注修正方向和幅度。上修太快 = 分析师在追逐事实而非预测事实。
陷阱 2b：把连续上调压扁成一个静态目标价
反例: 只写"Citi 目标价 100 港元，维持买入"。真正的信息是 65→90→100 的斜率，以及理由从估值倍数变成玻纤布 ASP、FR4 CCL 价格和毛利率上修。
规则: 目标价/EPS/ASP/GM 必须写 old_value → new_value → revision_magnitude_pct → driver_type。driver_type=earnings_path 时必须交给 Conductor 的 street_revision_delta，不能只放在研报摘要。
陷阱 3：把 GS 披露名单当成覆盖
反例: "Goldman Sachs 在 24 篇报告中提到 VeriSilicon"→ 误以为 GS 有覆盖。 真相: 全在"Company-specific regulatory disclosures"的覆盖公司名单中，无任何分析文字。GS 交易台有客户交易 → 研究部选择不覆盖 = 看见了但放弃。 规则: 必须区分：直接分析 > 行业提及 > 披露名单 > 无覆盖。披露名单=负面信号，不是正面。
陷阱 4：国际投行 KB 0 结果反复搜
反例: 搜了 5 轮确认 0 结果，浪费 15 分钟。 规则: 三路搜索（中文名+代码+英文名）0 结果后立即标注 [信息真空]，切东财 API 为主数据源。A 股中盘股国际投行覆盖率是结构性问题，不是操作失误。
好的分析长什么样（正例）
## EPS 预测可信度排序
| 卖方 | 2026E EPS | 修正轨迹 | 可信度 | 理由 |
|------|-----------|---------|--------|------|
| 东吴 | 0.60 | 0.39→0.5→0.6 (渐进) | 1st | 9篇跟踪最密，修正节奏合理 |
| 信达 | 0.61 | 0.05→0.16→0.61 (12x) | 5th | 8个月上修12x，在追逐而非预测 |
| 中邮 | 0.09 | 0.19→0.09 (下降) | 6th | 1月仍0.09，模型未纳入Q4订单 |
## 覆盖密度变化（顶部信号）
| 年份 | 机构数 | 报告数 |
|------|--------|--------|
| 2020 | 2 | 5 |
| 2024 | 4 | 5 |
| 2025 | 7 | 24 ← 暴增 |
| 2026 | 5 | 4(前4月) |
**诊断**: 覆盖密度 2025 年暴增 5x，与股价从 82→297 同步。A 股卖方覆盖密度与股价顶部高度正相关。
## Street Revision Delta
| 日期 | 机构 | 项目 | 原值 | 新值 | 幅度 | driver_type | 证据等级 |
|------|------|------|------|------|------|-------------|----------|
| 2026-06-04 | Citi | TP | HK$65 | HK$90 | +38% | earnings_path | S2摘要+价格数据待交叉 |
| 2026-06-09 | Citi | TP | HK$76 | HK$80 | +5% | earnings_path | S2摘要+ASP/GM假设 |
**诊断**: 连续上修若由 EPS/ASP/GM/订单驱动，是盈利路径变化；若只是换目标倍数，是 multiple_only。两者不能混为一个"目标价高"。
## 国际投行状态
| 机构 | 状态 | 证据 |
|------|------|------|
| GS | **披露但无报告** | 24篇披露附录列VeriSilicon，0分析 |
| MS/JPM/UBS/Nomura | 无覆盖 | 全文搜索0结果 |
| **诊断** | **看见但放弃** | 比不知道更负面 |
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.
- 5d ago First seen · 77 lines · 0 tokens per session scan A cac4b5b33788
report-agent is an agent published in the GitHub repository Howard-Jerry/quant-agent-skills (2 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,447 tokens. 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 agents, from other repositories
risk-analyst
Quantitative risk assessment across 5 FA-specific dimensions -- market/volatility, concentration, FX/currency, tax/account routing, fundamental/balance-sheet. Each scored 1-5. Produces overall risk score + mitigation actions. Adapted from TauricResearch/TradingAgents riskmgmt (aggressive/conservative/neutral…
portfolio-manager
Final synthesis judge -- receives all analyst + persona + technical + risk agent outputs, produces a decisive 5-level conviction rating (Buy/Overweight/Hold/Underweight/Sell) with account-routing, position-size guidance, and tax-aware action plan. Adapted from TauricResearch/TradingAgents portfoliomanager. Use after…
technical-analyst
Technical analysis -- price action, MA20/50/200 trend stack, RSI, MACD, Bollinger Bands, ATR, support/resistance levels. Uses Yahoo Finance MCP historical data. Adapted from TauricResearch/TradingAgents marketanalyst.
macro-analyst
Macro environment analyst -- snapshots rates/FX/commodities/indices/VIX, assesses cycle position, maps impact to a specific ticker or sector.
fundamentals-analyst
Deep financial statement analyst -- pulls 4+ years of income/balance/cashflow, computes trends, flags quality-of-earnings issues, outputs structured health card.
sec-filings-analyst
SEC filings analyst -- pulls 10-K, 10-Q, 8-K from SEC EDGAR (free, no API key), extracts Risk Factors, MD&A, and material events. For US-listed companies only.