mcp-redmine
MCP server for Redmine that provides full read/write coverage of issues, projects, wiki, time tracking, and more through 73 tools via the Redmine REST API.
README
mcp-redmine
A Model Context Protocol server that connects Claude (Desktop or Code) — or any other MCP client — to a Redmine instance through its REST API.
It exposes 73 tools covering, in practice, the entire stable Redmine REST API: issues, relations, projects, memberships, versions, categories, users, groups, roles, wiki, time tracking, attachments, project files, news, and search. Point your agent at a Redmine project and let it read and manage issues, log time, edit wiki pages, and administer users — all through natural conversation.
Features
- Full read/write coverage of issues, including bulk updates, relations, watchers, and comments.
- Project administration: create/update/archive projects, manage members, versions, and categories.
- User and group administration, with deliberately no "delete user" or "delete project" tool — see Security.
- Wiki, including nested pages and file attachments.
- Time tracking, attachments, project files, news, and full-text search.
- Readable errors: every failure comes back as a clear sentence pulled from Redmine's own validation response, not a raw stack trace.
- Nothing to run: a single Python process talking stdio to your MCP client — no database, no server to host, no Docker required.
Prerequisites
- Python 3.10+
- A Redmine instance with the REST API enabled:
- In Redmine, go to Administration → Settings → API and enable "Enable REST web service" (if not already on).
- Go to My account → API access key and copy the value.
Installation
With uv (recommended)
uv tool install mcp-redmine
With pip
pip install mcp-redmine
From source
git clone https://github.com/alsimoes/mcp-redmine.git
cd mcp-redmine
python3 -m venv venv
source venv/bin/activate # Linux/Mac
# venv\Scripts\activate # Windows
pip install -e .
Configuration
The server reads its configuration from two environment variables:
| Variable | Required | Description |
|---|---|---|
REDMINE_URL |
Yes | Base URL of your Redmine instance (no trailing slash). |
REDMINE_API_KEY |
Yes | Your API access key, from My account → API access key. |
REDMINE_TIMEOUT |
No | Per-request timeout in seconds. Defaults to 15. |
Claude Desktop
Edit (or create) the Claude Desktop configuration file:
- Mac:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"redmine": {
"command": "mcp-redmine",
"env": {
"REDMINE_URL": "https://redmine.example.com",
"REDMINE_API_KEY": "your_api_key_here"
}
}
}
}
If you installed from source instead of via uv tool install / pip install, point command at the interpreter in your virtual environment and
pass the script as an argument:
{
"mcpServers": {
"redmine": {
"command": "/full/path/to/mcp-redmine/venv/bin/python",
"args": ["-m", "mcp_redmine"],
"env": {
"REDMINE_URL": "https://redmine.example.com",
"REDMINE_API_KEY": "your_api_key_here"
}
}
}
}
Restart Claude Desktop afterwards. The tools icon (🔨) should show the Redmine tools as available.
Claude Code
claude mcp add redmine \
--env REDMINE_URL=https://redmine.example.com \
--env REDMINE_API_KEY=your_api_key_here \
-- mcp-redmine
Confirm with:
claude mcp list
Other MCP clients
Any client that can launch a local process and speak MCP over stdio works the
same way: run mcp-redmine (or python -m mcp_redmine) with REDMINE_URL
and REDMINE_API_KEY set in its environment.
Run locally (optional)
Useful for a quick sanity check before wiring the server into a client:
cp .env.example .env # then edit .env with your values
export $(grep -v '^#' .env | xargs) # or use a tool like direnv
mcp-redmine
If there's no connection error, the server is ready — it waits for MCP messages on stdin/stdout, so seeing nothing happen in the terminal is normal. Press Ctrl+C to stop it.
Usage examples
Once configured, just talk to the agent:
- "List the open issues assigned to me in the Website project."
- "Create a bug in Website titled 'Login button unresponsive on mobile', priority high."
- "Move issues #101, #104, and #110 to In Progress and assign them to Alice."
- "Log 3.5 hours on issue #204 for today, activity Development."
- "What's on the wiki page 'Deployment' for the Infra project?"
- "Search for 'timeout' across all projects."
Available tools
73 tools, grouped by resource. The full reference — signatures, parameters,
return shape, and the caveats that matter (like how precedes reschedules
issues, or that list_custom_fields requires an administrator) — lives in
docs/TOOLS.md.
| Resource | Tools |
|---|---|
| Issues | list_issues, get_issue, create_issue, update_issue, bulk_update_issues, delete_issue, add_watcher, remove_watcher, update_journal_note |
| Issue relations | list_issue_relations, create_issue_relation, delete_issue_relation, chain_issues |
| Projects | list_projects, get_project, create_project, update_project, archive_project |
| Project members | list_project_members, add_project_member, update_project_member, remove_project_member |
| Versions | list_project_versions, create_project_version, update_version, delete_version |
| Categories | list_project_categories, create_project_category, update_project_category, delete_project_category |
| Users | get_current_user, list_users, get_user, create_user, update_user, update_my_account |
| Groups | list_groups, get_group, create_group, update_group, delete_group, add_user_to_group, remove_user_from_group |
| Roles | list_roles, get_role |
| Wiki | list_wiki_pages, get_wiki_page, create_or_update_wiki_page, attach_file_to_wiki_page, delete_wiki_page |
| Time tracking | list_time_entries, get_time_entry, log_time, update_time_entry, delete_time_entry, list_time_entry_activities |
| Attachments | attach_file_to_issue, get_attachment, update_attachment, delete_attachment |
| Project files | list_project_files, upload_project_file |
| News | list_news, get_news_item, create_news, update_news, delete_news |
| Search | search, list_saved_queries |
| Metadata | list_statuses_and_priorities, list_trackers, list_custom_fields, list_document_categories |
API coverage
| Redmine API resource | Status |
|---|---|
| Issues | Complete (list, read, create, update, bulk, delete, watchers) |
| Issue relations | Complete |
| Projects | Complete minus deletion |
| Project members | Complete |
| Versions | Complete |
| Issue categories | Complete |
| Users | Complete minus deletion |
| Groups | Complete |
| Roles | Complete (the API is read-only) |
| Wiki | Complete, including nesting and attachments |
| Attachments | Complete |
| Project files | Complete |
| News | Complete |
| Time tracking | Complete |
| Search and saved queries | Complete |
| Trackers, statuses, priorities, activities, custom fields | Complete (the API is read-only) |
| Document categories | Complete (the API is read-only) |
The two omissions are deliberate. delete_project and delete_user are
irreversible and take other people's content down with them — issues, time
entries, wiki pages, authorship. archive_project and update_user with
status=3 cover the real use case reversibly. Deleting for real remains a web
UI operation, where confirmation is explicit and human. See
SECURITY.md for the full reasoning.
Trackers, issue statuses, priorities, and roles have no write operations in the Redmine API — they're configurable only in Administration. That's not a gap in this server.
Outside the stable REST API, and therefore outside this server's scope: repositories and changesets, wiki page protection, and instance-wide settings screens.
Error messages
Errors are handled centrally: when Redmine rejects a call, the server reads
the response body — that's where the real explanation lives, in the errors
field — and returns that sentence. Without a useful body, it falls back to
the HTTP status with a likely reason:
| Status | Message |
|---|---|
| 401 | invalid or missing API key |
| 403 | not allowed to perform this operation, or the module is disabled for this project |
| 404 | not found (check the ID, or this endpoint may not exist in this Redmine version) |
| 409 | conflict — the resource was changed by someone else |
| 422 | Redmine rejected the data during validation |
This applies to every tool, read and write alike.
Troubleshooting
- "REDMINE_URL and REDMINE_API_KEY must be set": the server checks its
configuration at startup and exits immediately with this message rather
than starting up broken. Check the
envblock in your MCP client's configuration. - Every call fails with a network error: confirm the server process can
actually reach
REDMINE_URL— MCP clients run the server as a local process, so it needs the same network access as your machine, not the browser. - 401 on every call: the API key is wrong, or "Enable REST web service" is off in Redmine's settings.
- 403 on a specific tool: usually a missing permission on the role of the
user who owns the API key.
get_roleshows exactly which permissions a role has. - A tool responds 404 that you'd expect to work: a few endpoints
(
update_journal_note,update_news/delete_news,update_attachment) were added in specific Redmine versions — see docs/TOOLS.md for the minimum version of each.
Security
See SECURITY.md for the full model. In short: an API key grants access equivalent to the user who owns it, so create a dedicated Redmine user with a role scoped to what you want the agent to do, and use that user's key — don't try to limit the agent by editing this server's code.
Contributing
See CONTRIBUTING.md.
License
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。