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 zhangxiaoqiang1991/luopan --skill companygit clone --depth 1 https://github.com/zhangxiaoqiang1991/luopanWrote 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/zhangxiaoqiang1991/luopan/company)<a href="https://agentmods.dev/skills/zhangxiaoqiang1991/luopan/company"><img src="https://agentmods.dev/badge/skills/zhangxiaoqiang1991/luopan/company/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/zhangxiaoqiang1991/luopan/company"><img src="https://agentmods.dev/badge/skills/zhangxiaoqiang1991/luopan/company.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 | $0.00089 | $0.02286 |
| Opus 5 | $0.00044 | $0.01143 |
| Sonnet 5 | $0.00018 | $0.00457 |
| Haiku 4.5 | $0.00009 | $0.00229 |
Grade A, and why
luopan-company-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 13d 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 — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
公司研究
目标
不输出公司百科，而是用可追溯事实回答两个相互独立的问题：
- 在当前价格和风险下，该公司是否值得继续作为投资对象研究；
- 对求职者而言，该公司是否值得继续投递，或某个具体业务/岗位/Offer 是否值得加入。
共享事实底座，但禁止将投资与求职合成一个总分。
交互原则
- 用户只提供公司名且没有用途线索时，只追问一道三选一：投资为主、求职为主、投资与求职并重；不要求其复述商业模式或填写问卷。
- 用户选择后直接研究。用途不明时不得自动猜测、不得默认双主线，也不得先做完再让用户纠正。投资与求职并重只在用户明确选择时使用。
- 求职信息不足时先给公司级或业务级结果，再邀请用户补充岗位和 Offer；不将补充信息设为首次使用门槛。
- 报告结尾给出 3–5 个基于当前结论的可继续追问方向。
工作流
1. 识别公司和研究模式
确认公司全称、品牌/法人/母子公司关系、上市状态、上市地、证券代码和研究基准日。法律注册地与实际经营总部不同时必须并列说明，例如“港交所上市主体，注册地为开曼群岛，实际经营总部位于深圳”；不得用注册地暗示主要经营地。只有同名实体会导致显著不同结果且无法自行确认时，才询问一个简短消歧问题。
将任务标记为：一般公司了解、投资、求职或双模式。读取并严格执行 references/audience-routing.md：已有明确意图时直接开始；用途不明时必须先让用户在投资为主、求职为主、投资与求职并重中选择。不得用公司属性、品牌印象、搜索热度或 Agent 评分代替用户选择。
2. 选择数据路线
读取 references/data-routing.md，按美股 SEC 申报主体、A 股、港股、其他上市市场或非上市公司路由。多地上市时明确本次证券标的和估值币种。
3. 建立共享事实底座
读取 references/evidence-standard.md 和 references/language-and-sources.md。搜集公司身份与披露质量、财务与资本配置、商业模式与单位经济、行业生态位与竞争优势、管理层与治理、组织/业务/岗位信号。默认输出中文：有同等权威的官方中文材料时优先中文；只有外文一手材料时中文转述并保留原始链接和原文标题，不为中文化牺牲证据等级。
对每条核心数据保留时间、币种、单位、口径、证据链接、页码/章节和是否为计算值。非上市公司无可验证披露时，禁止补全精确营收、利润和增速。
4. 运行判断引擎
- 投资或双模式：读取并执行 references/investment-framework.md。
- 求职或双模式：读取并执行 references/career-framework.md。
- 求职或双模式：同时读取 references/interview-questions.md，生成低防御的面试反问 10 问、温和追问及仅供用户后台复盘的绿/黄/红判定。不得把求职反问写成尽调审讯。
投资结论必须将公司质量与当前交易条件分开。求职结论必须标明它是公司级初筛、业务级判断还是 Offer 级判断。
5. 对抗验证
以最强反方视角检查：
- 是否把管理层叙事当成事实；
- 是否把相关当成因果；
- 是否混用财年、币种、会计口径或母子公司；
- 是否用单一毛利率、增速、估值或人员信号推出强结论；
- 是否存在能推翻当前判断的反证；
- 投资与求职结论是否被不当互相替代。
修正报告后再输出，不在最终报告中假装没有矛盾或信息局限。
6. 生成统一报告
读取 references/report-design.md，先写 30 秒决策卡、关键争议与折叠底稿。再读取 references/report-model.md，建立包含 facts、data_health、sections、sources 和 quiz_cards 的结构化 JSON 真源，并由同一份对象生成：
What ships with it
22 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/openai.yaml 285 B
- examples/example_report.json 2.5 KB
- references/audience-routing.md 3.0 KB
- references/automation-roadmap.md 1.2 KB
- references/career-framework.md 4.6 KB
- references/data-routing.md 6.5 KB
- references/evidence-standard.md 2.7 KB
- references/interview-questions.md 8.6 KB
- references/investment-framework.md 4.7 KB
- references/language-and-sources.md 3.0 KB
- references/price-safety-margin.md 3.4 KB
- references/quiz-cards.md 2.7 KB
- references/report-design.md 3.7 KB
- references/report-model.md 4.4 KB
- scripts/render_report.py 18 KB runs code
- scripts/sec_fetch.py 7.5 KB runs code
- tests/fixtures/sec_company_tickers.json 189 B
- tests/fixtures/sec_companyfacts.json 718 B
- tests/fixtures/sec_submissions_fpi.json 280 B
- tests/fixtures/sec_submissions.json 434 B
- tests/test_render_report.py 2.1 KB runs code
- tests/test_sec_fetch.py 2.7 KB runs code
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.
- 13d ago First seen · 133 lines · 89 tokens per session scan A 2e82534d3fa8
luopan-company-research is a skill published in the GitHub repository zhangxiaoqiang1991/luopan (385 stars, last pushed 1mo ago), licensed MIT. It adds 89 tokens to every session and 2,286 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 2026-08-30.
Other skills, from other repositories
tcx-strategy
Author and manage standalone Codex strategy skills named strategy- when the user wants an agent-readable investment strategy, library entry, entry or exit criteria, sizing rule, evidence standard, or decision-ready procedure. Start tool-using management with $tcx-strategy on the first meaningful line of a root turn…
tcx-brain
Author and manage TradingCodex Investment Brains. Use when the user asks to put principles or a Wiki insight into a Brain; create, inspect, revise, or validate a user-owned investment-brain- source; or install, update, activate, deactivate, roll back, or remove a managed Brain. Natural-language source authoring is…
tcx-memory
Review past investment decisions, run point-in-time historical replays, compare resolved forecasts and forward outcomes, and validate lesson candidates. Use when the user asks what happened before, why a thesis changed, whether a strategy lesson held out of sample, or what prior evidence is relevant to a current…
tcx-plan
Clarify an ambiguous TradingCodex mandate before immediate or recurring work by asking focused questions and preserving scope, constraints, action boundaries, approval posture, and stop conditions. Use when the user explicitly asks to plan, scope, or stress-test investment work, or when a missing choice in a schedule…
tcx-server
Operate TradingCodex server, read-only workspace viewer, MCP, update readiness, and broker connector status without build-turn or execution authority.
tcx-order-allow
Authorize at most one TradingCodex order submission or cancellation later in the current root native Codex turn. Use only when the user's first meaningful invocation is $tcx-order-allow with mode paper, validation, or live, including a saved Codex app Scheduled Task prompt.