Jobctl

What DmytroRybka/jobctl MCP server does

DmytroRybka/jobctl MCP server gives AI agents access to a structured job application tracker backed by jobctl.app. It is designed for recording and reviewing the operational details of a developer’s job search, including applications, pipeline status, priorities, salary information, contacts, interviews, notes, deadlines, analyses, and attached documents.

The service separates agent activity from human review. Agents write data through MCP tools, while people can use jobctl.app to view a pipeline board, analytics, interview timeline, and CV or cover-letter attachments. The repository is published in the MCP Registry as io.github.DmytroRybka/jobctl.

How it works

The recommended connection is a remote streamable HTTP MCP endpoint at https://app.jobctl.app/mcp. OAuth 2.1 with dynamic client registration is supported, and static bearer tokens beginning with jc_ can be used by clients that need token-based authentication. The account settings page at jobctl.app is where a token is created.

For clients that prefer local process communication, the repository includes a standalone stdio server named server.mjs. This bridge calls the jobctl.app API and exposes the same ten tools as the remote service. It can be run directly with Node.js or inside Docker. The server can list its tools without a token for introspection, but each tool call requires JOBCTL_TOKEN.

The API does not make server-side LLM calls; the connected agent supplies the AI behavior. Changes are recorded as auditable events with an actor label such as agent, human, or system.

Setup and configuration

For a remote setup, configure an MCP client with the streamable HTTP URL and either OAuth 2.1 or an authorization header containing a jobctl bearer token. Claude Code can add the endpoint with its HTTP transport and then install the repository’s Claude Code plugin or skill from the marketplace.

For local stdio use, clone the repository, install its npm dependencies, and start server.mjs with JOBCTL_TOKEN set. Docker is also supported: build the image and pass the token as an environment variable when starting the container. The supplied material does not specify additional required environment variables.

Tools and capabilities

DmytroRybka/jobctl MCP server exposes ten tools:

• add_application creates records from structured fields or parses a short raw description, with duplicate detection.
• list_applications filters records by status, priority, text, or saved views.
• get_application returns application details, including timelines, contacts, interviews, and attachments.
• update_application changes fields such as status, priority, salary, and next action.
• add_note appends information to an application timeline.
• log_interview records scheduled or completed interviews, attendees, and links.
• get_due returns overdue and upcoming next actions.
• get_planned presents an agenda for today, tomorrow, or the current week.
• attach_analysis stores one replaceable company-analysis Markdown file up to 100 kB.
• attach_file uploads CV or cover-letter files as base64 data.

Limitations and notes

The service stores data in the user’s account, with full JSON export and GDPR account deletion available according to the repository documentation. File and analysis handling has the stated format and size constraints. A valid token is required for operations, even though the local server supports unauthenticated tool discovery. DmytroRybka/jobctl MCP server is distributed under the MIT license.