vesta-mcp-server-savesearch

vesta-mcp-server-savesearch

MCP tool server for MLS saved searches that enables listing, proposing, saving, updating, and deleting a user's saved property searches.

Category
访问服务器

README

vesta-mcp-server-savesearch

MCP tool server for MLS saved searches — list, propose, save, update and delete a user's saved property searches.

The build order, the tests for each step and the reasoning behind them live in notes/saved-search-incremental-build-and-test-plan.md in the backend repo (smart-search-for-guestSide). This repo implements steps N1–N9 of that plan, which are phases 5, 7 and 9 of its build sequence. Read those three phases before writing code here. (That notes/ directory is untracked in the backend repo, so a fresh clone will not have the file — ask for a copy.)


Getting started

git clone https://github.com/<owner>/vesta-mcp-server-savesearch.git
cd vesta-mcp-server-savesearch
make setup

make setup needs only Python 3.11+ and uv. It installs dependencies, installs both git hooks, creates the secrets baseline, and verifies that ruff, mypy and pytest actually run. It is safe to re-run.

If uv is missing:

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows (there is no `sh` in cmd or PowerShell)
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

make itself is optional. Nothing in this repo's git hooks or CI depends on it — GNU Make is not on a stock Windows machine's PATH, so every make <target> below has an exact equivalent with no make involved:

python scripts/bootstrap.py                         # same as `make setup`
uv run --frozen python scripts/tasks.py <task>      # same as `make <task>`
uv run --frozen python scripts/tasks.py help        # every task, with descriptions

setup is the one thing that is not a tasks.py task: it has to run before a virtualenv exists, so it cannot go through uv run.

Daily commands

make check        # the three blocking CI jobs: fast hooks, mypy, tests + coverage
make test         # fast suite only
make format       # auto-fix lint and format
make audit        # report known CVEs in the locked dependency set
make help         # every task, with descriptions

check deliberately excludes the dependency audit, which is advisory in CI and needs network access. Run make audit when you want it.

You rarely need to run these by hand — the git hooks cover it.

What runs when

When What runs Roughly
git commit ruff lint, ruff format, secret scan, file hygiene, lock freshness, version-pin checks ~1s
git push mypy (strict, whole tree) + full test suite with the coverage gate seconds
pull request five parallel jobs: the fast hooks, mypy, tests, a Docker build, and an advisory CVE audit ~1 min

The split is deliberate. Commits stay fast so nobody reaches for --no-verify; the slow checks still run before code leaves your machine.

Tests marked contract or e2e hit live services, need real credentials, and never run in CI or in the hooks. Run them deliberately:

make test-contract   # -m contract
make test-e2e        # -m e2e

Layout

src/vesta_saved_search/   the package — all product code goes here
tests/                    will mirror src/ as it grows; also tests the tooling
scripts/                  developer tooling (bootstrap, task runner, consistency checks)

The src/ layout is load-bearing, not a style choice: it means every tool config targets src and tests and never a hand-maintained list of module names. See docs/reusing-this-setup.md.

Reusing this setup on another project

Every config file here is project-agnostic except pyproject.toml (five values) and the Dockerfile (its CMD module path and EXPOSE port). Copy the rest unchanged. See docs/reusing-this-setup.md for the exact list.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选
mcp-server-qdrant

mcp-server-qdrant

这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。

官方
精选
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选