← 返回 Skills 市场
caijichang212

Investment Research

作者 CaiJichang · GitHub ↗ · v0.2.1 · MIT-0
cross-platform ⚠ suspicious
373
总下载
0
收藏
3
当前安装
2
版本数
在 OpenClaw 中安装
/install investment-research
功能描述
提供公司或行业的全面投研分析,涵盖财务、行业格局、估值、技术面及风险催化,助力专业投资决策。
使用说明 (SKILL.md)

Investment Research(投研分析)

目标(Goal)

用"可复盘"的研究框架输出客观、可验证、带风险边界的投研结论;把"事实/数据"和"判断/假设"明确分开。

先问清楚(Intake)

在开写前,优先收集这些最少信息(缺失则在报告里标注假设):

  1. 标的(Ticker/市场/币种)与投资期限(短/中/长)
  2. 风险偏好与约束:是否可承受回撤、是否可用杠杆/期权
  3. 目标:择时交易还是长期配置?是否已有仓位、成本、计划加减仓
  4. 数据偏好:你提供财报/研报,还是我用公开信息检索(可能非实时),默认使用工具获取公开信息

工作流(Workflow)

Step 1 — 数据与事实层(Facts first)

  • 优先用:公司公告/财报、交易所披露、权威统计、主流券商一致预期(如可得)。
  • 获取数据工具:
    1. 推荐qveris-official:当需要股价、财报等结构化数据、专业财经数据或更强的工具聚合能力时使用。
    2. tavily-search:基本信息查询,搜索简单网页数据,并交叉验证,作为补充。
  • 输出时必须:
    • 给出引用来源(URL/机构/报告名)+ 数据日期/口径
    • 多源交叉验证(至少 2 个独立来源)
    • 不确定/无法验证:明确写"未知/待验证",不要脑补。

Step 2 — 基本面(Fundamental / 基本面)

  • 三表(资产负债表/利润表/现金流量表)联动看:增长、盈利质量、现金流、杠杆与偿债。
  • 拆商业模式与护城河(moat):客户是谁、价值主张、成本结构、议价能力、可复制性。
  • 找"反直觉"风险点:一次性项目、会计口径变化、应收/存货异常、资本开支压力。

Step 3 — 行业(Industry / 行业研究)

  • 明确行业口径与产业链位置;给 TAM/SAM/SOM(若无法量化则说明原因)。
  • 竞争格局:核心对手、份额变化、差异化、价格战可能性。
  • 政策/监管/地缘:对收入、成本、准入的影响路径。

Step 4 — 估值(Valuation / 估值)

  • 相对估值:PE/PB/PEG/EV-EBITDA 对比同行与历史分位(注意可比性与会计口径)。
  • 绝对估值:必要时给 DCF/情景区间(Bull/Base/Bear),把关键变量写清楚。
  • 输出估值区间优于单点目标价;注明数据日期与货币。

Step 5 — 技术面(Technical / 技术分析)

  • 只做"时点与风险管理"辅助:趋势(多周期)+ 关键位(支撑/阻力)+ 量价验证。
  • 指标作为证据而非结论:MA、MACD、RSI、KDJ、布林带等(见参考)。
  • 给可执行计划:入场区间、无效点/止损(stop-loss)、目标与跟踪规则。

Step 6 — 结论、催化剂、风险与反证

  • 催化剂(catalysts):未来 3–12 个月可验证事件 + 可能影响方向。
  • 风险:列 Top 3–7,并给"监控指标/触发条件"。
  • 反证(disconfirming evidence):什么发生会推翻你的核心观点。

输出规范(Output Standard)

  • 默认输出:一份《投研分析报告》+ 一段"行动清单"。
  • 明确区分:
    • 事实(Facts):带来源与时间
    • 假设(Assumptions):可被验证/证伪
    • 判断(Judgement):基于事实与假设
  • 避免确定性措辞:用"可能/大概率/条件成立时"。
  • 必须包含风险提示与免责声明。

模板与参考资料(Resources)

  • 生成报告时:优先按 references/report-template.md 的结构输出。
  • 指标口径不确定时:查 references/indicator-cheatsheet.md

快速示例(Prompts that should work)

  • "按基本面+行业+估值分析一下 XX(给 bull/base/bear)"
  • "把 XX 最近 3 年的财务质量拆开讲,看看有没有风险点"
  • "用技术面给一个交易计划:支撑阻力、止损止盈怎么设"
  • "对比 XX 和 YY:谁更值得配置?给关键分歧与跟踪指标"

工具要求(Tool Requirements)

推荐工具

  • qveris-official(首选):用于获取股价、财报等结构化数据和专业财经数据
  • tavily-search(备用):用于基本信息查询和网页数据补充

工具使用策略

  1. 优先使用结构化数据源(qveris-official)
  2. 交叉验证至少 2 个独立来源
  3. 明确标注数据来源、日期和口径
  4. 无法验证的数据明确标注"未知/待验证"
安全使用建议
What to check before installing or enabling this skill: - Metadata consistency: the package shows conflicting metadata (registry claimed no required env vars and version 0.2.1, but _meta.json/README/SKILL.md reference QVERIS_API_KEY, TAVILY_API_KEY and version 0.3.0 with a GitHub homepage). Ask the skill author or check the listed GitHub repo to confirm the canonical source and correct version. - API keys: the skill expects finance API keys (qveris/tavily). Only provide keys with limited scope and rotate them if you later revoke access. Do not reuse high-privilege or broadly-scoped secrets. - Verify tool endpoints: confirm 'qveris-official' and 'tavily-search' are legitimate trusted services and that the integration in your OpenClaw config points to official endpoints. If unsure, test with a throwaway or low-permission key first. - Least privilege: configure keys with read-only, query-limited scopes where possible and avoid embedding keys in files; follow the skill's own Security Advice (use env vars, .gitignore, rotate keys). - Operational testing: run the skill on a non-sensitive or public ticker first and verify returned source URLs, timestamps, and that the skill cites independent sources as it claims. - If you need higher assurance: request the canonical repository/source and changelog from the owner, or inspect any remote tool connectors your platform will install before granting credentials. Given the functional coherence but packaging/documentation mismatches, proceed only after resolving the metadata inconsistencies and confirming which env vars are actually required.
功能分析
Type: OpenClaw Skill Name: investment-research Version: 0.2.1 The investment-research skill bundle provides a structured framework for financial analysis, including fundamental, technical, and industry research. It utilizes legitimate external tools (qveris-official and tavily-search) for data retrieval and includes comprehensive documentation, security best practices for API key management, and a professional reporting template (SKILL.md, CONFIG.md, report-template.md). No evidence of malicious intent, data exfiltration, or harmful prompt injection was found.
能力评估
Purpose & Capability
The skill's stated purpose (structured investment research) aligns with the documented behavior: it instructs the agent to fetch financial data, analyze fundamentals/industry/valuation/technicals, and produce a report. However, there is an internal inconsistency: the top-level registry metadata shown to the evaluator lists no required env vars, while _meta.json and CONFIG.md explicitly reference QVERIS_API_KEY and TAVILY_API_KEY. Version/homepage fields also mismatch (registry shows version 0.2.1 and no homepage, while SKILL.md/README/_meta show 0.3.0 and a GitHub homepage). These mismatches reduce confidence in packaging quality.
Instruction Scope
SKILL.md is explicit and stays within the investment-research scope: it instructs using qveris-official and tavily-search to fetch public financial data, to cite sources and dates, to cross-validate at least two sources, and to separate facts/assumptions/judgements. It does not instruct reading unrelated local files or exfiltrating arbitrary system data.
Install Mechanism
This is an instruction-only skill with no install spec and no code files, minimizing on-disk execution risk. It does recommend configuring OpenClaw tools (qveris-official, tavily-search) in the platform config, which is normal for data-source integrations.
Credentials
Requiring API keys for financial data providers (QVERIS_API_KEY, TAVILY_API_KEY) is proportional to the skill's function. The problem is that the top-level 'Requirements' reported to the evaluator showed none, while _meta.json and CONFIG.md require these env vars—this inconsistency should be resolved before trusting the skill. Confirm which credentials are actually required and limit their scopes; do not provide broader secrets than necessary.
Persistence & Privilege
The skill does not request elevated platform privileges and 'always' is false. It asks the operator to enable/configure external tools in the OpenClaw config (tool-level config), which is expected. There is no evidence it modifies other skills or system-wide settings beyond its own recommended tool entries.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install investment-research
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /investment-research 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.2.1
Version 0.2.1 Changelog: - Updated version number to 0.3.0 in SKILL.md to reflect progress. - No functional changes to workflows, requirements, or usage guidance. - Documentation updates only: minor version increment and metadata refresh. - No changes to logic or output standards. All framework and analysis steps remain consistent.
v0.2.0
适合中国A股的投研分析技能,用于公司/股票/ETF/行业的专业分析
元数据
Slug investment-research
版本 0.2.1
许可证 MIT-0
累计安装 3
当前安装数 3
历史版本数 2
常见问题

Investment Research 是什么?

提供公司或行业的全面投研分析,涵盖财务、行业格局、估值、技术面及风险催化,助力专业投资决策。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 373 次。

如何安装 Investment Research?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install investment-research」即可一键安装,无需额外配置。

Investment Research 是免费的吗?

是的,Investment Research 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Investment Research 支持哪些平台?

Investment Research 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Investment Research?

由 CaiJichang(@caijichang212)开发并维护,当前版本 v0.2.1。

💬 留言讨论