
Ljg Invest
Turn founder materials or meeting notes into a depth investment-style report centered on whether the project is an order-creation machine—not a spreadsheet of multiples.
Overview
ljg-invest is an agent skill most often used in Validate (also Idea research) that produces investment-style reports asking whether a project is a compounding order-creation machine.
Install
npx skills add https://github.com/lijigang/ljg-skills --skill ljg-investWhat is this skill?
- Single organizing lens: wealth as illuminated order; invest for a better order generator
- Five-block report skeleton—这是什么, 秩序创造机器判定, and follow-ons—with depth where signal exists
- Flywheel, post-shock strength, and gravitational vs pushed resource questions
- Custom赛道 definition that穿透 surface labels (e.g. cognitive infrastructure vs “search company”)
- Pairs with Research skill or subagent for public names needing fresh financials
- 5 report blocks as flexible skeleton
- 3 machine angles: flywheel, post-shock, pull vs push resources
Adoption & trust: 3.4k installs on skills.sh; 5.5k GitHub stars; 2/3 security scanners passed (skills.sh audits).
What problem does it solve?
You have meeting notes or a deck but only conventional valuation chatter that does not tell you if the system actually compounds under competition and shocks.
Who is it for?
Indie operators diligencing startups, side-project acquisitions, or strategic bets when narrative materials are messy but rich.
Skip if: Retail ticker picks, compliance-grade securities filings, or builders who only need unit economics in a spreadsheet.
When should I use this skill?
User says 投资报告, 投资分析, 分析这个项目, investment report, invest analysis, or supplies entrepreneur conversation records for evaluation.
What do I get? / Deliverables
You receive a structured report with custom赛道 framing and a machine verdict on flywheel, resilience, and pull—so you can scope commitment, partnership, or pass.
- Dimensional summary table with custom赛道
- Order-creation machine assessment
- Investment-style narrative report sections as warranted by evidence
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Canonical shelf is Validate because the skill judges whether a venture’s structure can compound before you treat it as something to build, partner with, or fund. Scope fits the report’s heart: flywheel, shock resilience, and pull vs push growth—deciding what kind of machine you are actually buying into.
Where it fits
Compare two AI infrastructure plays using custom赛道 language before picking which problem to solve yourself.
Decide whether to join a startup as early hire based on flywheel and shock-history evidence in notes.
Sanity-check whether pricing power aligns with a pulling market vs pushed enterprise sales.
How it compares
Use instead of generic SWOT or DCF chat when you want a qualitative compounding-machine lens, not a template financial model.
Common Questions / FAQ
Who is ljg-invest for?
Solo founders, angel-adjacent builders, and agent users evaluating projects from pitches, BP snippets, or founder interview transcripts.
When should I use ljg-invest?
During validate scope when deciding invest/partner/build; during idea research when comparing opportunities; when the user asks for 投资报告, investment report, or invest analysis on pasted notes.
Is ljg-invest safe to install?
The skill may suggest fetching public data via other research skills; review the Security Audits panel on this page and vet any network subagents before use.
SKILL.md
READMESKILL.md - Ljg Invest
# 投资报告 生成一份投资分析报告。核心只问一个问题:这个东西在创造新秩序,还是在搬运旧秩序? ## 认知起点 财富不是钱,是被欲望照亮的秩序。投资就是拿手里的秩序去换一台更好的秩序生成器。 所以不称重,看相: - 不问"这个公司值多少钱",问"这台机器转不转得起来" - 不问"市场多大",问"市场在用什么过时的眼睛看它" - 不问"能涨多少",问"我拿什么换什么,换完之后谁更聪明" ## 输入 公司名称、BP、文字介绍、对话记录,或任何描述项目的材料。知名公司只给名字即可——用 Research skill 或 subagent 抓取最新财报和行业数据。 ## 报告结构 以下五个区块是骨架,不是填空题。对某个项目来说哪个区块最有料,那个区块就多写;没料的一两句带过或直接跳过。报告为判断服务,不为完整性服务。 ### 一、这是什么 一张表 + 一句自定义赛道定义。 | 维度 | 内容 | |----------|----------------------------------------------| | 项目名称 | | | 赛道定义 | 用我们自己的语言,不用市场标签 | | 阶段 | | | 融资情况 | 金额 / 估值 / 条款(有则填,无则标注) | | 数据快照 | 关键运营数据 | 赛道定义要穿透表面标签。"搜索引擎公司"是隔的,"人类认知基础设施的垄断运营商"是不隔的——后者告诉你它真正在做什么。 ### 二、秩序创造机器判定 这是整份报告的心脏。不逐项打分,而是回答一个问题:**这台机器转不转得起来?** 从三个角度看: **飞轮在不在转?** 系统有没有越用越好的结构——用户越多数据越多,数据越多产品越好,产品越好用户越多?这个循环是已经在转、刚开始、还是停着?转了多久?加速还是匀速?如果不转,什么卡住了? **冲击后变强还是变弱?** 竞争来了、技术变了、市场塌了——这台机器是碎掉、扛住、还是吃掉冲击变成自己的燃料?历史上有没有被冲击过?结果如何? **资源是被推过来的,还是自己来的?** 扩张靠的是一个一个谈、一块一块买(推),还是别人主动涌过来、因为不来就亏(引力)?有没有"不推而聚"的迹象? **综合判定**: - 秩序创造机器 — 飞轮在转,冲击后变强,资源自己来 - 有潜力 — 飞轮有结构但还没验证转起来 - 秩序搬运 — 把已有的东西重新排列,没有生成新秩序 ### 三、创生公式 每台秩序创造机器都有一个核心算法。用一句话写出来。 参考: - 亚马逊 = 利润→再投资→降成本→降价→更多用户→更多利润 - 特斯拉 = 硬件采数据→数据训练算法→算法重新定义硬件 - Google = 每次人类找答案的方式迁移时,成为新方式的默认基础设施 然后回答: - 这个公式被验证过几次?验证到什么程度? - 别人在跑相似的公式吗?差异在哪? ### 四、市场看见的 vs 我们看见的 这一节决定投资时机。 **它在 S 曲线的哪里?** 积累期、拐点、加速期、平台期。如果在拐点前——什么条件能触发拐点? **市场在用什么旧眼睛看它?** 市场给它贴的标签是什么?这个标签遮住了什么?我们的框架看到了什么市场看不到的?认知差有多大——这是超额收益的来源。 三个信号检测认知折价: - 需要很复杂的解释别人才能听懂? - 定价持续异常(部分之和 ≠ 整体)? - 已有的类比全部失效("像 X 但又不像 X")? **它控制了什么别人拿不走的东西?** 权力来源是什么——数据、分发、标准、网络效应?这种控制是静态的(品牌、专利)还是动态的(越变越强)? 这个稀缺性未来会不会位移?项目有没有能力跟着位移走? **它在搭哪趟便车?** 三种成本正在坍缩——理解成本、协作成本、行动成本。这个项目骑在哪一种上?坍缩释放的能量被它捕获了多少? ### 五、换不换 **交换建议**:建议投资 / 建议观察 / 建议放弃 **如果投资**:建议金额范围、关键条款 **核心假设**:这个决策依赖哪几个假设?每个假设附退出信号——什么数据出现说明假设错了,该跑。 **未解问题**:3-5 个对决策至关重要但还没答案的问题,按优先级排。 ### 最后一句 用一句话回答:这个项目的本质是什么?创造新秩序,还是搬运旧秩序? ## 输出 - 格式:org-mode - 目录:`~/Documents/notes/` - 命名:denote schema — `YYYYMMDDTHHMMSS==z--投资分析-PROJECT_NAME.org` - 例:`20260326153000==z--投资分析-example-ai.org` - 用 Write 工具写入,写完告知完整路径 ## 生成规则 1. 基于真实信息,不编造。信息不足的直接标注,不硬撑 2. 敢判断。"既可能好也可能坏"是废话,不许出现 3. 每个判断附证据——数据、引用、具体事实 4. 禁止出现:赛道很大、团队优秀、前景广阔、蓝海市场 5. 以说透为准,不计字数。两千字能说清就两千字,需要七千字就七千字 6. 中文撰写