Weave 0d69c60c

MCP文档知识

提供方 AaronBakerDev

文档知识 stdio · 本地进程 无需鉴权 社区 / 第三方
查看接入方式

Weave是一个ChatGPT原生的记忆平台,允许用户捕捉、组织和与朋友共享记忆。

一句话选型结论

按资料字段整理

Weave 0d69c60c 属于文档知识类 MCP 服务器,stdio · 本地进程,无需鉴权,有可直接执行的安装命令。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stdio · 本地进程 鉴权无需鉴权 部署方式本地部署
GitHub 星标 1 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

适用场景

记忆管理TypeScript本地部署协作工具ChatGPT集成知识管理共享平台

安装与配置

stdio · 本地进程 无需鉴权
客户端在本机拉起子进程,数据不出本机
python -m venv .venv

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

    stdio 服务在本地起进程,需要先装好 Python,再按安装命令拉取依赖。

  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。
  • 资料标注无需鉴权,装好就能连。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 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。
  • 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

  • 完全不想在本地装环境的人。
  • 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
Weave 0d69c60c当前 未标注 stdio · 本地进程 本地部署 有 2026-05-17
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AaronBakerDev上游标注的维护方 AaronBakerDev
分类文档知识本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式无需鉴权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 · Python本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

9.8 KB
来自项目仓库与上游目录的原文,未做删改

Weave: ChatGPT-Native Memory Platform

An explorable memory canvas inside ChatGPT. Capture, organize, and weave memories together with friends.

✨ Start with doc.md for the step-by-step practical build guide.

Quick Navigation

  • doc.md ← START HERE - Ordered, practical guide to ship v1 (Milestones A-D + tasks)
  • specs/product-spec.md - Product vision & UX
  • specs/technical-spec.md - Architecture & decisions
  • schema.sql - PostgreSQL schema with RLS policies

Current Status (2025-10-24)

✅ Platform pillars are in place
- ChatGPT UI (Next.js 14) implements the marketing home, inline Weaver card, memory detail editor, canvas view, search, and public/user routes.
- /api/mcp bridge translates ChatGPT tool calls into FastAPI requests using the manifest in chatgpt-mcp-manifest.json.
- FastAPI backend ships fully wired routers for memories, search, weave links, permissions, invites, artifacts, public slugs, follows, graph, export, plus health/rate limiting/JWT verification.
- Indexing worker listens to memory_event, rebuilds TSV and pgvector embeddings, and degrades gracefully if OPENAI_API_KEY is absent.

⏳ Outstanding work (tracked in ai-docs/sprint-2025-10-24.md)
- Upgrade local Node.js to ≥18.17 to satisfy Next.js runtime requirements and re-verify the Tailwind/PostCSS upgrade.
- Finish hybrid search parity by reintroducing the edge-boost term and queuing indexing for non-text layers.
- Refresh README/doc onboarding for Docker-less Postgres setups and updated UI/backend coverage.
- Harden automated testing (integration tests against live Postgres, Canvas interactions, etc.).

Immediate Focus

  1. Node.js Upgrade: Use nvm, asdf, or Volta to install Node ≥18.17, then re-run npm run build / npm run dev to confirm Tailwind/PostCSS fixes.
  2. Hybrid Search Polish: Implement the edge-boost component in services/api/app/routers/search.py and ensure layer appends enqueue indexing.
  3. Docs Refresh: Update doc.md/README onboarding for non-Docker Postgres flows and capture the current feature surface.
  4. Test Coverage: Run services/api/tests/test_api.py against a local Postgres (with pgvector) and document results; expand UI smoke tests as needed.

Milestones

Milestone Goal Acceptance
A - Capture & Recall Create/lock/search/detail view. Deployed. P0 endpoints working, search <2s p95, end-to-end works
B - Weaving & Canvas Link memories, Canvas with richer cards, zoom-to-enter Create weave, see thread, smooth nav
C - Sharing & Invites Role-based access, invite flow, public groundwork Invite user, accept, append layer, slug resolves
D - Public Mirror Follows, public profiles, optional events Public pages <500ms p95 cached

Directory Structure```

weave/
├── doc.md ← BUILD GUIDE (start here!)
├── schema.sql # PostgreSQL schema + RLS
├── README.md # (this file)
│
├── services/api/ # Python FastAPI backend
│ ├── app/
│ │ ├── main.py
│ │ ├── models.py # SQLAlchemy ORM + Pydantic models
│ │ ├── deps.py # JWT auth + RLS-aware sessions
│ │ ├── routers/
│ │ │ ├── memories.py # CRUD, layers, permissions, suggestions
│ │ │ ├── search.py # Hybrid search (edge boost pending)
│ │ │ ├── weave.py # Memory connections
│ │ │ ├── invites.py # Invite create/accept
│ │ │ ├── follows.py # Follow graph + feeds
│ │ │ ├── public.py # Public slugs
│ │ │ ├── graph.py # Canvas graph feed
│ │ │ └── artifacts.py # Upload/download proxies
│ │ ├── workers/
│ │ │ └── indexing.py # Embedding + TSV refresh worker
│ │ └── db/
│ │ ├── migrations/ # SQL files (run in order)
│ │ └── rls.sql
│ ├── requirements.txt
│ ├── .env.example
│ └── docker-compose.yml
│
├── apps/
│ ├── chatgpt-ui/ # Primary Next.js ChatGPT App
│ │ ├── app/
│ │ │ ├── api/
│ │ │ │ ├── mcp/ # JSON-RPC bridge to FastAPI
│ │ │ │ └── proxy/ # REST proxy to FastAPI
│ │ │ ├── canvas/ # Canvas experience
│ │ │ ├── memory/ # Memory detail route
│ │ │ ├── search/ # Search UI
│ │ │ ├── p/, u/, a/ # Public slugs, author pages, artifact redirect
│ │ │ └── globals.css # Tailwind entry (v4)
│ │ └── package.json
│ │
│ └── next-app/ # Earlier stub preserved for reference
│
├── specs/
│ ├── product-spec.md # Vision, UX, features
│ ├── technical-spec.md # Stack, architecture
│ └── api-docs.md # API reference (TODO)
│
└── docs/
├── DEPLOY-RAILWAY.md # Railway deployment guide
└── ...


## Tech Stack

| Component | Choice | Notes |
|-----------|--------|-------|
| **Frontend** | Next.js 14 (Vercel starter) | ChatGPT Apps SDK |
| **UI Bridge** | Node.js MCP shim | `/api/mcp` route proxies to Python |
| **Backend** | Python 3.11 FastAPI | Railway container |
| **Database** | PostgreSQL 14+ + pgvector | RLS policies for multi-tenancy |
| **Search** | Hybrid scoring | 0.55 cosine + 0.35 BM25 + 0.10 edge boost |
| **Embeddings** | OpenAI `text-embedding-3-large` | 1536 dimensions |
| **Storage** | Backblaze B2 | S3-compatible, 10GB free, videos/images via signed URLs |
| **Auth** | JWT | JWKS URL-based verification |
| **Deployment** | Vercel + Railway | Fully managed, no DevOps |

## Local Development

### One-Command Dev

If you have Docker, Python 3.11+, and Node ≥18.17 installed, you can start everything (Postgres + migrations + FastAPI + Next.js) with a single command from the repo root:```
make dev

This launches:
- Postgres with pgvector on port 5432 (container name: weave-pg)
- FastAPI on http://localhost:8000
- Next.js UI on http://localhost:3000

Stop the app processes with Ctrl+C. The database container stays running for faster restarts. To stop it later:```
make down && docker stop weave-pg


### 1. Start PostgreSQL + Backend```bash
cd services/api

# Run migrations in order (PostgreSQL must be running and pgvector enabled)
psql &quot;$DATABASE_URL&quot; -f app/db/migrations/0001_init.sql
psql &quot;$DATABASE_URL&quot; -f app/db/migrations/0002_idempotency.sql
psql &quot;$DATABASE_URL&quot; -f app/db/migrations/0003_core_locked_at.sql
psql &quot;$DATABASE_URL&quot; -f app/db/migrations/0004_memory_event.sql
psql &quot;$DATABASE_URL&quot; -f app/db/rls.sql

# Install and run FastAPI
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python -m uvicorn app.main:app --reload --port 8000

> No Docker? Use a local Postgres 14+ instance with the vector extension installed. Update DATABASE_URL (for example postgresql://localhost:5432/weave) before running the migrations above.

2. Start Indexing Worker (optional)```bash

cd services/api
source .venv/bin/activate
python app/workers/indexing.py


### 3. Start ChatGPT UI```bash
cd apps/chatgpt-ui  # or next-app
npm install
PYTHON_API_BASE=http://localhost:8000 npm run dev

4. Use the app locally

  • Create a memory on http://localhost:3000 with the Weaver card.
  • Open the Canvas at http://localhost:3000/canvas and click nodes to enter detail.
  • Public pages live under /p/{slug} and author pages under /u/{handle}.

Key Implementation Details (from doc.md)

Endpoint Pattern:
- Routes: /v1/* prefix
- Request/response: include query + results: [{memory, score, reasons}]
- Idempotency: via idempotency-key headers on create/lock

RLS & Auth:
- Every request sets SET LOCAL app.user_id = &#x27;&#x27; (see deps.py)
- Policies enforce: owner sees all, contributors see + write, viewers see only, public = anyone

Search Scoring:hybrid_score = 0.55 * cosine_similarity + 0.35 * ts_rank + 0.10 * edge_boost
- cosine_similarity: Vector distance on narrative embedding
- ts_rank: PostgreSQL full-text search (TSV column)
- edge_boost: +0.1 if memory connected to query context

Indexing Trigger:
- On core lock and text/reflection layer append
- Build document: title + core.narrative + [text_layers] + [captions]
- Update tsv and embedding columns

Deployment

See deployment guides:
- DEPLOY.md - Quick deployment guide (Render + Vercel)
- RENDER-SETUP.md - Detailed Render backend setup
- VERCEL-SETUP.md - Vercel frontend setup
- BACKBLAZE-B2-SETUP.md - Backblaze B2 storage (videos/images)
- DOMAIN-SETUP.md - Custom domain configuration

Key env vars:
- DATABASE_URL (Render PostgreSQL)
- JWT_JWKS_URL (Clerk auth)
- OPENAI_API_KEY (embeddings)
- AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY (Backblaze B2)
- S3_ENDPOINT_URL (Backblaze B2 endpoint)
- ALLOWED_ORIGINS (CORS)

Next Steps

→ READ doc.md FOR THE BUILD GUIDE

Then:
1. Check existing code in services/api/
2. Start with task #1: JWT verification in deps.py
3. Follow tasks in order (they unblock each other)
4. Test locally before each deploy

References

  • PostgreSQL RLS: https://www.postgresql.org/docs/current/ddl-rowsecurity.html
  • Vercel ChatGPT Apps: https://vercel.com/docs/ai
  • Railway Docs: https://docs.railway.app
  • OpenAI Embeddings: https://platform.openai.com/docs/models/text-embedding
  • FastAPI: https://fastapi.tiangolo.com

常见问题

7 问
答案由本站依据条目字段整理,不引用上游原文
Weave 0d69c60c 怎么安装?

在客户端按下面的安装命令接入:python -m venv .venv。装完在客户端配置里引用同一个包或端点。

Weave 0d69c60c 怎么在客户端里配置?

它的传输方式未标注,在客户端 MCP 配置里按官方文档填写启动命令和参数,保存后重启客户端生效。

Weave 0d69c60c 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。

Weave 0d69c60c 需要 API Key 吗?

资料标注为无需认证,装好就能连;如果实际调用时报鉴权错误,按官方文档补上凭据。

Weave 0d69c60c 能提供哪些能力?

资料显示能力标签包括 记忆管理、TypeScript、协作工具、知识管理。具体工具清单以官方文档为准。

Weave 0d69c60c 需要什么运行环境?

运行时为 Python;属于本地进程,需要在本机装好运行环境。

Weave 0d69c60c 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

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 2026-05-17。