Are you the author? Sign in to claim
Local MCP (stdio) server for Redmine. Work with issues, projects and time entries from any MCP client.
A local Model Context Protocol server (stdio) for Redmine. Point an MCP client at it (Cursor, VS Code, Claude, Codex, Windsurf, Cline, Zed, JetBrains AI and others) and work with Redmine issues, projects and time entries straight from the editor.
| Tool | What it does | Key parameters |
|---|---|---|
list_issues | List/filter issues, paginated | project_id, status_id, assigned_to_id, query, limit, offset, all |
get_issue | Full details of one issue + comments/history + attachments | id, include_journals (default true) |
list_projects | Accessible projects | limit |
get_project | Project details, versions and members | id |
create_issue | Create an issue | project_id, subject, description, tracker_id, priority_id, assigned_to_id, status_id |
update_issue | Update issue fields | id + any of subject, status_id, priority_id, assigned_to_id, done_ratio, notes |
add_comment | Add a note to an issue | id, notes |
log_time | Log a time entry on an issue/project | issue_id or project_id, hours, activity_id, comments, spent_on |
list_time_entries | List time entries, filtered | issue_id, project_id, user_id, from, to, limit, offset |
list_enumerations | IDs of trackers/statuses/priorities/activities | none |
search | Full-text search (issues, wiki, news, …) | q, project_id, types, titles_only, open_issues, limit, offset |
get_current_user | Account behind the API key (id, login, name) | none |
Every tool ships MCP tool annotations (readOnlyHint, destructiveHint, …), so clients can auto-approve read-only calls.
No manual install needed. The client launches the server on demand with npx, which
downloads it on first run and caches it after. Just add the config for your client under
Connect.
Prefer to run from a local clone? See Development.
My account > API access key > Show.
⚠️ Never commit your API key. Copy
.env.exampleto.envfor local use;.envis gitignored.
The server reads REDMINE_URL and REDMINE_API_KEY from the environment. Provide them via the client's MCP config env block.
Edit ~/.cursor/mcp.json (global) or .cursor/mcp.json in a project root:
{
"mcpServers": {
"redmine": {
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
Reload Cursor and enable the server in Settings > MCP.
Edit .vscode/mcp.json in your workspace (or your user mcp.json). Note the top-level key is servers, not mcpServers:
{
"servers": {
"redmine": {
"type": "stdio",
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
claude mcp add redmine \
--env REDMINE_URL=https://redmine.your-company.com \
--env REDMINE_API_KEY=your_key \
-- npx -y redmine-mcp-stdio
Or edit ~/.claude.json / the project .mcp.json directly (same shape as Claude Desktop below).
Edit claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"redmine": {
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
Restart Claude Desktop afterwards.
Add a server table to ~/.codex/config.toml:
[mcp_servers.redmine]
command = "npx"
args = ["-y", "redmine-mcp-stdio"]
env = { REDMINE_URL = "https://redmine.your-company.com", REDMINE_API_KEY = "your_key" }
If node is installed via nvm, use the full path from which npx as command.
Edit ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"redmine": {
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
In Cascade, open MCP servers > Manage > View raw config, paste, then refresh. Supports ${env:VAR} interpolation if you'd rather not inline the key.
In the Cline pane, click the MCP Servers icon and choose Configure MCP Servers to open cline_mcp_settings.json:
{
"mcpServers": {
"redmine": {
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
Zed calls MCP servers context servers. Open the command palette and run zed: open settings, then add to settings.json (note "source": "custom"):
{
"context_servers": {
"redmine": {
"source": "custom",
"command": "npx",
"args": ["-y", "redmine-mcp-stdio"],
"env": {
"REDMINE_URL": "https://redmine.your-company.com",
"REDMINE_API_KEY": "your_key"
}
}
}
}
Zed restarts the server on save, so no editor reload is needed.
Settings > Tools > AI Assistant > Model Context Protocol (MCP) > Add:
npx (if node is installed via nvm, use the full path from which npx)-y redmine-mcp-stdioREDMINE_URL=https://redmine.your-company.comREDMINE_API_KEY=your_keyEnable "Automatically enable new and changed MCP servers", then Apply. Click the status icon and you should see 12 tools. Invoke them in chat with /.
REDMINE_URL=https://redmine.your-company.com REDMINE_API_KEY=your_key npx -y redmine-mcp-stdio
The process should start and wait silently on stdio (Ctrl+C to exit).
Failed to connect / server won't start: the command must resolve in the client's environment. If node is installed via nvm, a bare npx may not be found. Use the full path from which npx as the command.Missing REDMINE_URL or REDMINE_API_KEY: the env block wasn't passed. Double-check it's in the MCP config, not your shell.Redmine 401 / 403: bad API key, or the key's user lacks permission for that action.Redmine 404: the issue/project id doesn't exist, or REDMINE_URL points at the wrong host.| Variable | Required | Default | Description |
|---|---|---|---|
REDMINE_URL | yes | n/a | Base URL of your Redmine instance (http/https) |
REDMINE_API_KEY | yes | n/a | Personal API access key |
REDMINE_TIMEOUT_MS | no | 10000 | Per-request timeout in milliseconds |
REDMINE_MAX_RETRIES | no | 2 | Retries on 429/5xx with exponential backoff |
The code is organized in layers, with a single HTTP client injected into each tool
group so that tools never touch transport or process.env directly:
src/
index.ts entry point: load config, build client, register tools, serve stdio
config.ts env parsing + validation (Zod) into a typed RedmineConfig
version.ts single source of the version string
redmine/
client.ts RedmineClient: auth, timeout, retry, host pinning, secret redaction
errors.ts RedmineError / ConfigError + HTTP error normalization
paths.ts safe path-segment encoding + project-id validation
types.ts Redmine API response types
tools/
register.ts wires every tool group onto the server
issues.ts list/get/create/update issues + add_comment
projects.ts list/get projects (versions, members)
time-entries.ts log_time, list_time_entries
enumerations.ts trackers / statuses / priorities / activities
search.ts full-text search across issues, wiki, news, ...
users.ts get_current_user
helpers.ts guard(), ok()/fail(), ToolResult
format.ts pure presentation helpers
npm run build # compile src/ to build/
npm run typecheck # tsc --noEmit (strict)
npm run lint # eslint
npm run format # prettier --write
npm test # build, then run the vitest suite
npm run check # typecheck + lint + format:check + test (what CI runs)
See CHANGELOG.md for releases.
The API key is never logged and is redacted from error output; requests are pinned to the configured host, path parameters are encoded, and every call is bounded by a timeout and a response-size cap. Details and reporting: SECURITY.md.
MIT. See LICENSE.
Run analytics queries on ClickHouse — explore schemas, execute SQL, fetch results
Run Claude Code as an MCP server so any agent can delegate coding tasks to it
Browser automation using accessibility snapshots instead of screenshots
Google's universal MCP server supporting PostgreSQL, MySQL, MongoDB, Redis, and 10+ databases