Worklogger MCP

Use Jira worklogs, issue updates, and Bitbucket pull requests from an AI assistant. Worklogger runs a local MCP server and lets you decide which provider tools the assistant can call.

It is useful for three everyday jobs:

• check this week's worklogs and find assigned issues without time;
• read an issue, prepare a comment or field update, then review the change; and
• prepare a pull request with its reviewers and branch settings before creating it.

Every provider write returns a preview and requires a matching confirmation request. Worklogger stores credentials locally: Windows uses Credential Manager and Linux uses a private, permission-restricted user store. Tokens are never written to generated JSON or client configuration.

The public installer supports Windows x64 and Linux x64 with glibc 2.35 or newer, including Ubuntu 22.04+ and WSL2. It contains the native server for the current system and installs it in a versioned user location:

• Windows: %LOCALAPPDATA%\Worklogger\MCP\<version>;
• Linux/WSL: $XDG_DATA_HOME/worklogger/MCP/<version> or ~/.local/share/worklogger/MCP/<version>.

Node.js 18 or newer is required only to launch the public installer; Node is not needed while the server is running.

Guided setup

Run the installer in the same environment as the client: PowerShell for Windows applications, or the relevant WSL terminal for Codex or Claude Code in WSL.

npx @santiv343/worklogger

If this scope was previously configured for GitHub Packages, return to the public npm registry once:

npm config delete @santiv343:registry

Without arguments, Worklogger opens a guided terminal interface. It shows what is configured and missing, validates each provider as you connect it, and lets you install the server, clients, and optional assistant skills separately.

After setup, restart the registered client and try a read-only request such as:

Show my worklogs for this week.

To use a shared team profile:

npx @santiv343/worklogger setup --profile /path/organization.json

The profile is validated and installed in the user's configuration directory. It may define branding, modules, scopes, limits, and allowed capabilities, but never accounts or secrets. The wizard stores every person's token in Windows Credential Manager or a Linux atomic store protected by 0700/0600; no token is copied to JSON or MCP clients.

Manual or headless setup

Copy a reviewed mcp.json to a private location. Start from the read-only Jira example in the repository, enter any required token in the same shell without putting it in shell history, then run:

read -rsp "Jira API token: " WORKLOGGER_JIRA_API_TOKEN; echo
export WORKLOGGER_JIRA_API_TOKEN
npx --yes @santiv343/worklogger install \
  --config /private/path/mcp.json \
  --clients codex \
  --yes
unset WORKLOGGER_JIRA_API_TOKEN

The first --yes accepts npm's install prompt. The final --yes confirms Worklogger changes. --clients accepts all, codex, claude-code, claude-desktop, cursor, windsurf, qwen-code, gemini-cli, kiro, github-copilot, and trae-code. See the MCP client support guide for the target configuration used by each client. Add --skills only when workflow skills should be installed for every compatible assistant destination detected on that account.

The command never replaces unrelated or invalid client registrations. Restart registered clients, run npx @santiv343/worklogger status, and make one read-only request through the client to verify the connection.

Other commands

npx @santiv343/worklogger clients
npx @santiv343/worklogger skills
npx @santiv343/worklogger status
npx @santiv343/worklogger uninstall

uninstall removes only registrations owned by Worklogger, its MCP configuration, and its MCP credential. It does not touch Desktop credentials, unrelated integrations, Jira worklogs, pull requests, or versioned binaries that may still be running.

Learn more

• Install and use Worklogger
• Supported MCP clients
• Guide for assistants
• Permissions and privacy