AIO Tests MCP Server

AIO Tests MCP Server

Enables searching, reading, creating, and updating AIO Tests test cases in Jira from any MCP client, supporting Jira Cloud and Server/Data Center with a read-only mode.

Category
访问服务器

README

<!-- mcp-name: io.github.AmirhosseinHanifehzadeh/aio-tests-mcp -->

AIO Tests MCP Server

CI License: MIT Python 3.10+ MCP

An open-source MCP server for AIO Tests — test management for Jira.

Connect Claude, Claude Code, Cursor, VS Code, Windsurf or any other MCP client to your Jira test cases. Search and read them, inspect a project's field schema, browse the folder tree, and create or update cases — Classic and BDD — in plain conversation.

Works with Jira Cloud and Jira Server / Data Center. Self-hosted, so your credentials and test data never leave your infrastructure.

uvx aio-tests-mcp-server

Not on PyPI yet. The uvx command above goes live with the first release. Until then, use Running from source — it takes two commands.

Why this exists. The official AIO Tests MCP server is a closed-source, remotely hosted service with a tenant-specific URL and Cloud-only support. This one is self-hostable and auditable — you can read every line, fork it, and run it inside your own network. It is, as far as I know, the only AIO Tests MCP server that supports Jira Server / Data Center, and it has a read-only mode so you can safely point an AI at a production instance.


Tools

Ten tools, all scoped to a Jira project.

Tool Access What it does
aio_get_project read Confirm AIO Tests is enabled for a project and get its ID
aio_get_test_case_schema read Fields, custom fields, required fields and allowed values
aio_search_test_cases read Filter by title, key, folder, status, priority, type, tag, owner, covered Jira issue, automation status/key, and creation/update dates
aio_get_test_case read Full case detail including every step
aio_get_test_case_versions read Every saved version of a case
aio_create_test_case write Create a case with Classic or BDD steps
aio_update_test_case write Update fields, steps, metadata, priority, status or tags
aio_get_folder_hierarchy read Folder tree (or a flat list of full paths) for test cases, cycles or sets
aio_create_folder write Create a folder, creating missing parents along a path
aio_get_tags read Every tag configured for a project

Lookup fields accept names or IDs. You can say "Critical" instead of 4, or /Regression/Checkout instead of a folder ID — values are resolved against the project configuration, so the model works in the same vocabulary you see in the AIO Tests UI.

Updates preserve formatting. Case updates round-trip the document with rich text enabled, so the formatting of fields you did not touch survives, and plain-text input is escaped to match.


Quick start

1. Get an access token

Jira Cloud — in Jira, open AIO Tests → (?) icon → API Access Token and generate one. The token identifies your tenant, so no URL is needed.

Jira Server/Data Center — the AIO Tests API is served from your Jira base URL and reuses your Jira credentials (a Personal Access Token, or username + password).

2. Add it to your MCP client

<details open> <summary><b>Claude Desktop / Claude Code</b> — Jira Cloud</summary>

{
  "mcpServers": {
    "aio-tests": {
      "command": "uvx",
      "args": ["aio-tests-mcp-server"],
      "env": {
        "AIO_API_TOKEN": "your_aio_access_token"
      }
    }
  }
}

</details>

<details> <summary><b>Claude Desktop / Claude Code</b> — Jira Server / Data Center</summary>

{
  "mcpServers": {
    "aio-tests": {
      "command": "uvx",
      "args": ["aio-tests-mcp-server"],
      "env": {
        "AIO_ENABLED": "true",
        "JIRA_URL": "https://jira.your-company.com",
        "JIRA_PERSONAL_TOKEN": "your_personal_access_token"
      }
    }
  }
}

</details>

<details> <summary><b>Cursor / Windsurf / VS Code</b></summary>

Same JSON as above, in the client's MCP settings file (~/.cursor/mcp.json for Cursor, .vscode/mcp.json for VS Code). Cursor also accepts a project-local mcp.json. </details>

In Claude Code you can add it in one command:

claude mcp add aio-tests --env AIO_API_TOKEN=your_token -- uvx aio-tests-mcp-server

3. Ask for something

"Search PROJ for published test cases tagged smoke in /Regression/Checkout, and show me the steps of the first one."

"Read PROJ-142 and write BDD test cases covering the acceptance criteria."


Configuration

Every setting is an environment variable; see .env.example for the full list. The essentials:

Variable Purpose
AIO_API_TOKEN AIO Tests access token (Cloud). Setting this alone enables the server.
AIO_ENABLED Set to true to enable Server/Data Center mode.
JIRA_URL Jira base URL (Server/DC).
JIRA_PERSONAL_TOKEN Jira PAT (Server/DC), or use JIRA_USERNAME + JIRA_API_TOKEN.
READ_ONLY_MODE true hides every write tool — recommended against production.
ENABLED_TOOLS Comma-separated allowlist of tool names.
AIO_URL Override the API base URL.
AIO_SSL_VERIFY false to skip certificate verification (testing only).

Proxy settings (AIO_HTTP_PROXY, AIO_HTTPS_PROXY, AIO_SOCKS_PROXY, AIO_NO_PROXY), client certificates and custom headers are supported too.

The same options are available as CLI flags:

uvx aio-tests-mcp-server --help

Read-only mode

Point an AI at a production Jira and you probably want it looking, not touching:

uvx aio-tests-mcp-server --read-only

Write tools are then hidden from the tool list entirely, so the model never sees them.


HTTP transport

For shared or containerised deployments, run it over Streamable HTTP instead of stdio:

uvx aio-tests-mcp-server --transport streamable-http --port 8000

The server exposes /mcp and a /healthz endpoint for Kubernetes probes.

Multi-tenant use

One server instance can serve several users. Each client sends its own AIO Tests token per request, and the request is scoped to that tenant:

{
  "mcpServers": {
    "aio-tests": {
      "url": "https://aio-mcp.your-company.com/mcp",
      "headers": {
        "X-Aio-Api-Token": "the_users_own_token"
      }
    }
  }
}

Authorization: Token <token> works as an equivalent to the X-Aio-Api-Token header.


Running from source

git clone https://github.com/AmirhosseinHanifehzadeh/aio-tests-mcp.git
cd aio-tests-mcp
uv sync
cp .env.example .env   # then fill it in
uv run aio-tests-mcp-server

Run the tests:

uv run pytest

Limitations

  • Folder rename, move and delete are not implemented. The AIO Tests public API exposes no endpoint for them.
  • Test cycles and test runs are not covered yet. The current tool set is scoped to test case management — cases, folders, tags and project schema. Contributions welcome.

FAQ

Is this the official AIO Tests MCP server? No. AIO Tests publishes its own MCP server as a hosted service — closed source, with a tenant-specific URL, and Cloud only. This is an independent, open-source implementation you run yourself. It is not affiliated with or endorsed by AIO Tests or Atlassian.

Does it work with Jira Data Center or Jira Server? Yes. Set AIO_ENABLED=true and JIRA_URL, and authenticate with a Jira Personal Access Token or username and password. On Server/DC the AIO Tests API is served from your Jira base URL, so it reuses your Jira credentials. This is the main reason to pick this server over the alternatives — the hosted ones are Cloud-only.

Can I use it with Claude Code, Claude Desktop, Cursor or VS Code? Yes — any MCP client works. See Add it to your MCP client for ready-to-paste config.

Is it safe to point at production Jira? Run it with --read-only (or READ_ONLY_MODE=true). Write tools are then filtered out of the tool list entirely, so the model never sees that creating or updating is an option.

Where do my credentials go? Nowhere but your own machine and your Jira instance. The server runs locally over stdio by default and talks straight to the AIO Tests REST API. There is no intermediary service.

Can one server instance serve a whole team? Yes — run it over HTTP and have each client send its own token in the X-Aio-Api-Token header. See Multi-tenant use.

Does it support test cycles, executions or attachments? Not yet. The current tool set covers test case management — cases, folders, tags and project schema. Cycles and executions are the obvious next step; see Limitations, and issues are welcome.


Contributing

Issues and pull requests are welcome — see CONTRIBUTING.md. The test suite covers the client, config resolution, every mixin and every tool; please keep it green and add cases for new behaviour.

License

MIT.

Portions of this project are derived from mcp-atlassian by @sooperset, also MIT-licensed. Thanks for the foundation.

This project is not affiliated with or endorsed by AIO Tests or Atlassian.

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
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
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
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
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
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
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
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