CEREBRO 战略情报引擎——扫描、分析并连接不同折扣仓库的知识
概览
扫描、评分，并在Markdown保险库中连接知识。像战略家一样思考，而不是搜索引擎。 CEREBRO读取结构化的降价库，对战略价值进行评分，检测时间敏感信号，绘制实体间关系图，并展示可操作的情报——包括你未曾察觉的联系。 它不是一个搜索工具。搜索回答“那是哪里？”CEREBRO回答“现在最重要的是什么，我还缺少什么？” 每一项发现都来自你实际写的链接、标签和前言。没有任何信息是凭散文统计来猜测的。 这个限制正是关键所在。一个从你的文字中推断意义的工具会产生自信的胡言乱语，而你无法在不重读每条笔记的情况下分辨真实发现和捏造的发现——这违背了目的。每个CEREBRO发现都可以通过打开两个文件来验证。 在 Windows PowerShell 上，先用 .venv\Scripts\Activate.ps1 激活，然后再运行 python -m pip install ..
概述（AI 翻译）
扫描、评分并连接知识，跨越一个Markdown保险库。 要像战略家一样思考，而不是搜索引擎。
版本： 3.0.1 ·许可： MIT ·Python： 3.11+
它的作用
CEREBRO读取了一个结构化的折扣金库，在战略上评分每一个细节 值，检测时间敏感信号，映射实体间的关系， 并揭示可操作的情报——包括你未曾发现的关联 知道要注意。
它不是一个搜索工具。搜索答案是“那是哪里？”CEREBRO 回答 “现在最重要的是什么，我错过了什么？”
| 它找到了 | 含义 |
|---|---|
| 未解决的实体 | 你反复链接它。你从未写下来。 |
| 隐式联系 | 两个笔记共享上下文，且从未被链接。 |
| 二度关系 | A连接B，B连接C。大脑表面AC↔。 |
| 瓶颈 | 有一点是所有无关工作领域都依赖的。 |
| 覆盖薄弱 | 标签只贴过一次——是意图，不是有记录的区域。 |
| 紧急信号 | 截止日期和关键词分类：紧急/高/标准/过去。 |
发现是结构性的，而非推断
本页由 AI 翻译，技术细节请以原文为准。
README
以下为原文 README 节选，完整内容请查看原文。
CEREBRO — Strategic Intelligence Engine
Scan, score, and connect knowledge across a markdown vault. Think like a strategist, not a search engine.
Version: 3.0.1 · License: MIT · Python: 3.11+
What It Does
CEREBRO reads a structured markdown vault, scores every note on strategic value, detects time-sensitive signals, maps relationships between entities, and surfaces actionable intelligence — including connections you did not know to look for.
It is not a search tool. Search answers “where is that?” CEREBRO answers “what matters most right now, and what am I missing?”
| It finds | Meaning |
|---|---|
| Unresolved entities | You link to it repeatedly. You never wrote it down. |
| Implicit connections | Two notes share context and were never linked. |
| Second-degree relationships | A links to B, B links to C. CEREBRO surfaces A↔C. |
| Bottlenecks | One note that unrelated areas of work all depend on. |
| Thin coverage | A tag applied once — an intention, not a documented area. |
| Urgency signals | Deadlines and keywords classified URGENT / HIGH / STANDARD / PAST. |
Findings are structural, not inferred
Every finding is derived from links, tags and frontmatter you actually wrote. Nothing is guessed from prose statistics.
That constraint is the point. A tool that infers meaning from your writing produces confident nonsense, and you cannot tell a real finding from a fabricated one without re-reading every note — which defeats the purpose. Every CEREBRO finding can be verified by opening two files.
Install
git clone https://github.com/microhustlestack/cerebro-skill.git
cd cerebro-skill
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install .
cerebro --version
On Windows PowerShell, activate with .venv\Scripts\Activate.ps1 before
running python -m pip install ..
Or deploy as an agent skill at the same time:
bash install.sh # package + Claude; detected Hermes/OpenClaw
bash install.sh package claude # install the CLI and one skill target
Explicit targets are package, claude, hermes, and openclaw. Supplying
targets replaces the defaults, so include package when you also want the
cerebro command installed.
Dependencies: Python 3.11+ and PyYAML. Nothing else.
Quick Start
# Full scan with gap analysis
cerebro ~/Documents/Obsidian/my-vault --gaps --report cerebro_report.md
# JSON index for an agent to reason over, gap findings included
cerebro ~/vault output/vault-index.json --gaps --quiet
# Report straight to stdout
cerebro ~/vault --quiet --report -
| Flag | Effect |
|---|
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add microhustlestack/cerebro-skill