devflow-mcp

devflow-mcp

A read-only MCP server that securely serves Jira (Server/Data Center) data to Claude Code without exposing the personal access token, with optional GitLab integration.

Category
访问服务器

README

devflow-mcp

Local, read-only MCP server that securely serves Jira (Server/Data Center) data to Claude Code — without ever exposing your PAT/token to the model. (GitLab tools + the /devflow skill land in later phases; see the plan.)

Plan: plans/260625-1353-jira-ai-dev-mcp/

Why

The token lives only inside this server's HTTP client (read from an env var). Claude calls tools like jira_get_ticket_bundle("PROJ-123") and gets back trimmed JSON — it never sees the credential, and the credential is never a tool argument.

Requirements

  • Python ≥ 3.10
  • A Jira Server/DC Personal Access Token (Bearer). Profile → Personal Access Tokens.

Setup

cd devflow-mcp
python3 -m venv .venv
source .venv/bin/activate
pip install -e .            # or: pip install -e ".[dev]" for tests

cp .env.example .env        # then edit JIRA_BASE_URL + JIRA_PAT

.env (gitignored) for local runs:

JIRA_BASE_URL=https://jira.your-company.com
JIRA_PAT=xxxxxxxxxxxx

Smoke test (MCP Inspector)

fastmcp dev server.py
# open the Inspector URL, call jira_get_ticket_bundle with a real issue key

First, call jira_get_fields once and note the custom field IDs for Sprint / Epic Link / Story Points — needed in later phases.

Register in Claude Code

Add to your project .mcp.json (commit this — it has NO secrets, only ${VAR} refs):

{
  "mcpServers": {
    "devflow": {
      "command": "/ABS/PATH/devflow-mcp/.venv/bin/python",
      "args": ["/ABS/PATH/devflow-mcp/server.py"],
      "env": {
        "JIRA_BASE_URL": "${JIRA_BASE_URL}",
        "JIRA_PAT": "${JIRA_PAT}",
        "GITLAB_URL": "${GITLAB_URL}",
        "GITLAB_TOKEN": "${GITLAB_TOKEN}"
      }
    }
  }
}

Put the real values in your shell (~/.zshrc). GitLab is optional — omit it and GitLab tools self-disable:

export JIRA_BASE_URL="https://jira.your-company.com"
export JIRA_PAT="xxxxxxxxxxxx"
export GITLAB_URL="https://gitlab.your-company.com"   # self-hosted; /api/v4 appended automatically
export GITLAB_TOKEN="glpat-xxxxx"                       # scopes: read_api, read_repository

Restart Claude Code, then the tools appear.

Tools (21, all read-only)

Jira core

Tool Purpose
jira_get_ticket_bundle(key) primary: issue + subtasks + links + comments + attachment metadata in one call
jira_get_issue(key, fields?) single issue, trimmed
jira_get_comments(key, limit?) recent comments
jira_get_subtasks(key) / jira_get_linked_issues(key) subtasks / linked issues
jira_search(jql, fields?, limit?) JQL search (paginated, capped)
jira_get_fields() resolve custom field IDs (setup)

Jira agilejira_get_epic, jira_get_epic_issues, jira_get_sprint, jira_get_sprint_issues

Jira attachments (image analysis)jira_get_attachments(key), jira_download_attachment(key, attachment_id) → downloads to a local path (token hidden) so Claude can read mockups/screenshots with vision.

GitLab (self-hosted, optional) — gitlab_search_mrs(ticket_key), gitlab_get_mr, gitlab_get_mr_commits, gitlab_get_commits, gitlab_get_commit_diff, gitlab_get_pipeline_status, gitlab_search_code, gitlab_get_file.

Security notes

  • Secrets are read from env only; never accepted as args, never returned, masked in errors.
  • Read-only: no write/transition/comment endpoints are implemented.
  • JIRA_VERIFY_SSL=false only for trusted internal self-signed certs.

Troubleshooting

  • Exits "Missing required environment variable" → set JIRA_BASE_URL / JIRA_PAT.
  • 401/403 → check PAT validity + project permissions.
  • SSL error on internal host → set JIRA_VERIFY_SSL=false (trusted hosts only).

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