PersonalKnowHow

MCP文档知识

提供方 未标注

文档知识 HTTP · 远程 未标注 社区 / 第三方 Claude DesktopClaudeCursorVS Code
查看接入方式

将您的LinkedIn/GitHub/课程历史转化为可通过MCP查询的知识图谱,使用真正的语义搜索而非关键词匹配。

一句话选型结论

按资料字段整理

PersonalKnowHow 属于文档知识类 MCP 服务器,HTTP · 远程,鉴权方式未标注,可直接接入 Claude Desktop、Claude、Cursor 等,有可直接复制的远程端点配置。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类文档知识 传输方式HTTP · 远程 鉴权未标注 部署方式远程部署
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

文档知识
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

记忆

安装与配置

HTTP · 远程 未标注
连远程 HTTPS 端点,通常为 Streamable HTTP
{ "mcpServers": { "personalknowhow-demo": { "command": "npx", "args": ["mcp-remote", "https://personalknowhow-demo.kxtwrdzt6g.workers.dev/mcp"] } } }

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    确认你的客户端支持远程端点,并拿到服务地址。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
文档检索问答

在你的文档库里按问题找答案。

知识库读写

把新资料写进知识库,随用随取。

内容整理

把零散资料归纳成结构化笔记。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude Desktop、Claude、Cursor,换客户端不用重新研究配置。

先想清楚的地方

  • 远程端点依赖网络与服务方可用性,注意密钥与调用额度。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在文档知识场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
PersonalKnowHow当前 未标注 HTTP · 远程 远程部署 有 未标注
dify 未标注 stdio · 本地进程 本地部署 有 2026-04-11
open-webui 未标注 stdio · 本地进程 本地部署 有 2026-04-11
JeecgBoot 未标注 stdio · 本地进程 本地部署 有 2026-04-11
Cherry Studio 未标注 未标注 未标注 无 2026-05-17
mempalace 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

文档知识
字段来自项目仓库与公开索引,中文说明由本站标注
提供方未标注上游标注的维护方 未标注
分类文档知识本站按上游分类展示
连接方式HTTP · 远程连远程 HTTPS 端点,通常为 Streamable HTTP
鉴权方式未标注远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · 未标注本地安装时用到的包标识与运行时
部署类型远程部署上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

9.9 KB
来自项目仓库与上游目录的原文,未做删改

Join the Waitlist · Live Demo · Issues

Turns a scattered personal learning/work history — LinkedIn, GitHub, course platforms, Gmail
completion emails, sibling project repos — into a unified knowledge graph, queryable with
semantic search instead of keyword matching. Run it locally against your own data in two
commands, no account required — or query the real, deployed
MCP servers described further down.

Asking Claude (via the deployed MCP server) to evaluate a job posting against the graph — real evidence pulled from LinkedIn/GitHub history, not a guess. (full-resolution video)

Try it locally in 60 seconds

No Cloudflare account, no signup, nothing deployed — just your own machine.```bash
git clone https://github.com/Georgi-Petkov/personalknowhow.git
cd personalknowhow
python quickstart.py
python ingest/query_local.py "what do I know about X"


`quickstart.py` auto-detects whatever sources are already available on your machine and skips
the rest with a clear reason — at minimum, an already-authenticated GitHub CLI (`gh auth login`)
or your sibling project repos are enough to get real results. `query_local.py` embeds your graph
with a small local model (`BAAI/bge-small-en-v1.5`
via `sentence-transformers`, downloaded once from Hugging Face on first run — the one real network
dependency of local mode, distinct from needing a *cloud account*) and ranks results by cosine
similarity, the same approach the deployed MCP servers use.

**Want your own LinkedIn history in the graph, not just GitHub/local-project evidence?** Request
your export at linkedin.com → Settings & Privacy → Data privacy → *Get a copy of your data*, then:```bash
python quickstart.py --linkedin ~/Downloads/LinkedInDataExport.zip
python ingest/query_local.py "what do I know about X"

quickstart.py handles unzipping and routing it to the right ingest script itself — no manual
file placement, no flags to figure out. (Requesting the export happens entirely on LinkedIn's
site and can take a few minutes for them to prepare — everything after that is the two commands
above.)

Every other source (edX, DataCamp, Gmail) needs its own one-time setup (a hand-populated JSON
file or OAuth credentials) — quickstart.py detects and skips each one gracefully with a
one-line reason if it's not set up; see ingest/CLAUDE.md for per-source
details if you want to add one.

Try the live demo

https://personalknowhow-demo.kxtwrdzt6g.workers.dev/mcp is a real, deployed MCP server — but
it's not a webpage. Opening that URL in a browser sends a plain GET, and MCP servers only
speak POST with JSON-RPC framing, so you'll just see a bare {"error":{"message":"Method not allowed."}}. That's expected, not broken — it means you're looking at it the wrong way.

The actual way to use it is as an MCP connector. In Claude Desktop, edit
claude_desktop_config.json (config file location):```json
{
"mcpServers": {
"personalknowhow-demo": {
"command": "npx",
"args": ["mcp-remote", "https://personalknowhow-demo.kxtwrdzt6g.workers.dev/mcp"]
}
}
}


Restart Claude Desktop, then ask something like *"use personalknowhow-demo to check if I have
Django experience"* — Claude calls the `query_knowhow` tool over MCP and gets back
semantically-matched evidence (courses, projects, certifications) with similarity scores, no
auth required.

If you just want to confirm the server is alive without setting up a client:```bash
curl -s https://personalknowhow-demo.kxtwrdzt6g.workers.dev/mcp \
  -X POST -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}}}'

A 200 with a JSON-RPC response back confirms it's live — the Accept header above is required;
without it the server correctly returns 406 Not Acceptable, which is a different, also-expected
error from the browser-GET one above.


Why this exists

Course-completion lists and keyword-matched resumes are a weak signal of what someone actually
knows. This project builds a real knowledge graph from primary sources (not self-reported
summaries), embeds every entry with a real embedding model, and exposes it as a queryable MCP
tool — so "do I have Django experience?" gets answered by walking real evidence (a project's
README, a course completion, an endorsement) with a similarity score attached, not a guess.

Architecture```

ingest/ Source-specific scripts → common schema
{title, type, provider, date, description, domain_tags}
corpus/ Generated markdown, one subfolder per source (not tracked — see Privacy below)
graph.json Extracted nodes/edges from corpus/ (not tracked)
mcp/ Public MCP server (Cloudflare Worker) — semantic search, no auth
mcp-private/ Private MCP server — same search, bearer-token gated, adds
signal-only evidence (job applications, career interests)


**Ingestion sources**: LinkedIn (via the Member Data Portability
API, EU-only —
see `docs/linkedin-connector-notes.md` for notes on the
manual-export alternative for other regions), GitHub (via the `gh` CLI, excluding forks — a fork
is evidence of browsing, not building), DataCamp/edX/Skilljar course completions, Gmail
(completion emails from other platforms), and sibling project repos (auto-discovered, evidenced
via README + tracked filenames + a keyword pass, not self-reported).

`ingest/merge.py` deduplicates across sources (idempotent — safe to re-run). `ingest/build_graph.py`
extracts nodes/edges from `corpus/` frontmatter into `graph.json`.

## The RAG layer

Both `mcp/` and `mcp-private/` are stateless Cloudflare Workers (`createMcpHandler`, no Durable
Object) that embed every corpus entry with Workers AI (`@cf/baai/bge-base-en-v1.5`, 768-dim) at
export time, and embed the query string at request time, then rank by cosine similarity. Two MCP
tools are exposed: `query_knowhow(topic)` for semantic search, and `list_by_type(type)` for a
plain listing. The private server additionally tags every result with an `evidence_tier`
(`demonstrated` vs. `signal_only`), so a job application or career-interest entry can never be
mistaken for proof of a skill.

## Privacy design

`corpus/` and `graph.json` are never public — no public-facing code reads them directly. The
**only** sanctioned public data source is `mcp/public_entries.json`, built by
`ingest/build_public_export.py` via a **fail-closed allowlist**: only explicitly listed corpus
categories (courses, projects, certifications, education, endorsements, positions, profile,
recommendations, articles) get exported. A new corpus category is excluded by default until
someone deliberately adds it to the allowlist — the same discipline that keeps job applications
and career-interest data out of the public server entirely; that data only exists in
`mcp-private/`, gated behind a bearer token, and is never committed to this repo either (see
`.gitignore`).

## Career-agent tooling

A second layer built on top of the same corpus: `ingest/analyze_job_postings.py` scores scraped
job postings against known skill coverage using the same embeddings (graded `known`/`peripheral`
similarity, not binary keyword matching), `ingest/cv_tailor.py` matches a posting's requirements
against CV bullets with an explicit two-tier system (exact-term matches vs. semantically-related
matches, the latter always labeled "verify before claiming" rather than asserted), and
`ingest/recommend_courses.py` cross-references course catalogs against coverage gaps.

## Running individual ingest sources manually

`python quickstart.py` (see the top of this README) runs everything below automatically for
whichever sources it detects. For finer control — a single source, non-default flags, or
re-running just one step after a corpus change — run any of these directly:```bash
pip install -r requirements.txt

# Run a specific ingest source, e.g.:
python ingest/github_ingest.py
python ingest/linkedin_api_ingest.py --domains PROFILE,POSITIONS,SKILLS

# Deduplicate corpus after any ingest run
python ingest/merge.py

# Build graph.json from corpus/
python ingest/build_graph.py

Each ingest/*_ingest.py script is independent — run whichever sources apply to you. All of them
write markdown into corpus// using the shared schema below.

Corpus schema

Every markdown file in corpus/ uses this YAML frontmatter:```yaml

title: "Advanced Python Programming"
type: "course" # course | certification | position | project | education | ...
provider: "LinkedIn Learning"
date: "2024-01-15"
description: "Free-text summary."
domain_tags:
- python
- programming



## Deploy your own hosted MCP server (optional, needs a Cloudflare account)

Local mode (above) is enough to query your own graph — this section is only for hosting it as a
real MCP server other clients/people can connect to, the same way the live demo works.```bash
cd mcp && npm install && npm run deploy        # public server
cd mcp-private && npm install && npm run deploy # private server
cd mcp-private && npm run secret                # set PRIVATE_MCP_TOKEN

Both need a Cloudflare account with Workers AI access ([ai] binding, remote = true in
wrangler.toml). Rebuilding the embeddings after a corpus change:bash CLOUDFLARE_ACCOUNT_ID=... CLOUDFLARE_AI_TOKEN=... python ingest/build_public_export.py CLOUDFLARE_ACCOUNT_ID=... CLOUDFLARE_AI_TOKEN=... python ingest/build_private_export.py

Adding a new ingest source

  1. Create ingest/_ingest.py that reads the raw export and writes markdown files to
    corpus// using the schema above.
  2. merge.py and build_graph.py require no changes — they scan corpus/ generically.
  3. If the new category should ever be public, add it deliberately to ALLOWLIST in
    ingest/build_public_export.py — it's excluded by default otherwise.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
PersonalKnowHow 怎么安装?

在客户端按下面的安装命令接入:{ "mcpServers": { "personalknowhow-demo": { "command": "npx", "args": ["mcp-remote", "https://personalknowhow-demo.kxtwrdzt6g.workers.dev/mcp"] } } }。装完在客户端配置里引用同一个包或端点。

PersonalKnowHow 怎么在 Claude Desktop、Claude、Cursor 里配置?

它的传输方式是HTTP · 远程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

PersonalKnowHow 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

PersonalKnowHow 需要 API Key 吗?

资料没有标注鉴权方式,是否需要密钥取决于服务本身,接入前按官方文档确认。

PersonalKnowHow 支持哪些客户端?

资料标注支持 Claude Desktop、Claude、Cursor、VS Code、Codex、Zed;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

PersonalKnowHow 能提供哪些能力?

资料显示能力标签包括 记忆。具体工具清单以官方文档为准。

PersonalKnowHow 需要什么运行环境?

属于远程端点,不需要在本地装运行时。

PersonalKnowHow 是本地还是远程?

它是远程端点,连服务地址即可。传输方式为HTTP · 远程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 未标注。