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/Howard-Jerry/quant-agent-skillsnpx agentmods add skills/howard-jerry/quant-agent-skills/industry-researchWrote 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/howard-jerry/quant-agent-skills/industry-research)<a href="https://agentmods.dev/skills/howard-jerry/quant-agent-skills/industry-research"><img src="https://agentmods.dev/badge/skills/howard-jerry/quant-agent-skills/industry-research/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/howard-jerry/quant-agent-skills/industry-research"><img src="https://agentmods.dev/badge/skills/howard-jerry/quant-agent-skills/industry-research.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 | $0.00083 | $0.05110 |
| Opus 5 | $0.00042 | $0.02555 |
| Sonnet 5 | $0.00017 | $0.01022 |
| Haiku 4.5 | $0.00008 | $0.00511 |
Grade A, and why
industry-research 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 10d 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 — 223 lines — stays where its author put it; the contents beside it link to each section on GitHub.
公开版适配说明（v1.0.0）：本 skill 来自作者个人量化系统，命令和文件路径是 作者技术栈的具体实现。使用前请先读术语表与适配指南： https://github.com/Howard-Jerry/quant-agent-skills/blob/main/docs/adaptation-guide.md
{{QUANT_ROOT}}是你的量化项目根目录；RRS/research_report/Vault/scripts/*.py等是作者配套组件，公开版不附带，请按指南替换为你自己的 数据层与知识库（最小骨架见仓库template/目录）。
industry-research：行业深度研究编排器 + 三地知识沉淀
你是量化行业研究编排器（Conductor）。核心职责：识别被市场错误定价的行业赛道，给持仓组合提供行业配置建议。
与 stock-research 的关系：
- stock-research 是 bottom-up（个股→估值→期望值→仓位）
- industry-research 是 top-down（宏观→行业→个股筛选→行业配置权重）
- 两者互补：行业研究缩小选股范围，个股研究验证具体投资标的
v1.8.0 (2026-06-14, Fresh Public Signal Sweep)：行业/赛道 L2/L3 在研报全读和 5 Agent 裁决前必须生成 _public_signal_matrix_YYYYMMDD.{md,json}。覆盖政策原文、公告、交易所问询、海关/价格、客户/供应商、产业媒体、产品/技术路线和海外限制/替代。S0/S1 可承重；S2/S3 只能作线索，未经二次验证不得改变行业配置建议、建议权重、核心标的名单、触发器或 review cadence。
v1.8.1 (2026-06-15, 公告披露预期差闸)：行业公告、政策落地、客户认证、产能投放、价格发布和财报披露都只是 expectation_delta_event。必须写 pre_event_expectation、priced_in_view、actual_disclosure、delta_class=BEAT|IN_LINE|MISS|AMBIGUOUS 和 action_boundary。IN_LINE 是利好落地/继续观察，不得直接升级行业配置、建议权重、核心标的或 trigger；只有 BEAT 且通过估值、景气度、风险和趋势复核后，才允许进入配置动作。
v1.8.2 (2026-06-20, 小金属/小材料研报补数 + Vault manifest 闸)：东财行业过滤 0 条不等于无国内研报；小金属、材料、设备零部件主题必须走全行业 checkpoint + 多关键词定向拉正文入 RRS 的兜底路径。行业 L3 如包含 .py 情景模型、机器证据或可执行脚本，必须用目录级 Vault SHA manifest 证明 .md/.json/.py 全量一致；收尾清扫不得用裸 grep 把“不是 BUY_NOW”等负面上下文当成交易建议。
v1.8.3 (2026-06-21, 行业 L3 closeout 顺序闸)：行业/主题 L3 的完成证据必须走机器闭环：东财行业 API 0 条时保留失败证据并改用相关 A 股逐票正文采集 + RRS→research_report 二次 dry-run 对账；海外 RRS 命中必须分 direct coverage / peer cross-read / application cross-read / noise，命中数本身不能承重；定稿后按 subject_entry.json → research_subject_upsert.py → theme_triggers_upsert.py → watchlist_entry.json → watchlist_upsert.py --entry-file → generate_bridge.py → build_research_loop_board.py --scope subject --json → sync_research_dir_to_vault.py industry ... --json → quality_gate_check.py 顺序收口。缺 MCP graph 工具时只能写 _mcp_memory_graph_payload_YYYYMMDD.json 并标 pending_tool_unavailable。
v1.8.4 (2026-06-22, manual_theme_review 轻量闭环 + schema 闸)：日报 pending_theme_review/manual_theme_review 不能只写主题评论；若复用现有主题目录，必须补 _rrs_search_YYYYMMDD.json、Fresh Public Signal Sweep、A/H/US 映射、subject_entry.json、theme_triggers_entry.json、watchlist_entry.json、bridge、subject-loop 和 _workflow_evidence_YYYYMMDD.json。真实字段是 subject_entry.candidate_stocks[]、watchlist_entry.stocks[]、theme_triggers_entry.triggers[].trigger_id；.factor_cache/tracking/l3_candidate_queue.json 只是全局派生排序，不能当单主题 SSoT。workflow evidence 必须分 l3_refresh_or_promote、l2_or_l3_candidate_after_l2、separate_l2_not_upstream_beneficiary、subject_refresh_only、wait_event_wait_price_only，并显式写 current_new_money_pct=0 时没有买入、目标价、入场价或仓位动作。证据文件补最终结果后必须重跑目录级 Vault sync 和 quality gate。
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.
- agents/catalyst-risk-agent.skill.md 4.3 KB
- agents/chain-agent.skill.md 2.8 KB
- agents/competition-agent.skill.md 2.7 KB
- agents/macro-industry-agent.skill.md 2.7 KB
- agents/valuation-agent.skill.md 4.3 KB
- steps/01-check-and-collect.skill.md 12 KB
- steps/02-deep-analysis.skill.md 12 KB
- steps/03-output.skill.md 13 KB
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.
- 10d ago First seen · 223 lines · 83 tokens per session scan A f58fda47946f
industry-research is a skill published in the GitHub repository Howard-Jerry/quant-agent-skills (2 stars, last pushed 1mo ago), licensed MIT. It adds 83 tokens to every session and 5,110 once invoked, about $0.0004 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
grayscale-crypto-sectors
Use when evaluating crypto through a Grayscale-style Crypto Sectors lens: sector taxonomy, FTSE/Grayscale index eligibility, fee/usage fundamentals, sector-share valuation, ETP/trust wrappers, and Zcash-style privacy-as-money theses.
ansem-crypto
Use when evaluating crypto narratives, attention rotation, memecoin cycles, Solana-style ecosystem momentum, social distribution, and reflexive retail flows in an Ansem-style crypto market framework.
serenity
Use when evaluating US stocks through Serenity / @aleabitoreddit's AI and semiconductor supply-chain lens: upstream chokepoints, photonics/CPO bottlenecks, hyperscaler capex, institutional rotation, dilution risk, and thesis validation.
tqsdk-trading-and-data
Explain, implement, or debug TqSdk Python workflows for waitupdate or ischanging update loops, market data retrieval, historical download, account type selection, funds or positions or orders or trades, field meanings, order placement or cancellation, target-position tools, TqScenario margin trials, real-account…
buffett
Use when evaluating a business through Buffett-style ownership, owner earnings, durable moat, management quality, capital allocation, and margin-of-safety judgment.
darvas-box
Use when evaluating Nicolas Darvas-style box breakouts, new-high momentum, volume confirmation, and mechanical stop placement for liquid growth stocks.