jobctl is a job application tracking tool for developers, with the core idea being "AI agents handle the organizing, humans handle the reviewing and planning." Through MCP (Model Context Protocol), it hands the writing of job application records to AI agents, and leaves reading and planning to a clean dashboard interface.
Main capabilities:
- Agent writes: AI agents write job search data through a remote MCP server, whose address is https://app.jobctl.app/mcp，使用可流式 HTTP transport and OAuth 2.1 authentication.
- Human reads: humans view the pipeline board, analytics data, interview timeline, and CV/cover letter attachments on jobctl.app.
- What can be recorded: job applications, interviews, status, salary, and next steps can be recorded (per the official description).
- Published to the official MCP Registry, identified as io.github.DmytroRybka/jobctl, and listed on Glama as a connector (status Healthy, tool quality A).
How to use:
- Claude Code plugin: the repository publishes a Claude Code plugin/skill. After installation, you can directly converse in natural language, for example "I applied to Bosch as Tech Lead, prio A, due friday", and Claude will keep the tracker up to date during the user's interviews.
- Other MCP clients:
- Remote method (recommended): any client that supports streamable HTTP can connect to https://app.jobctl.app/mcp。支持 OAuth 2.1 dynamic client registration; a static bearer token (starting with jc_) works on all clients.
- Local stdio server: the repository includes server.mjs, which is a standalone MCP server (stdio) that exposes the same 10 tools through the jobctl.app API. This server can start and list tools (introspection) without a token; every tool call requires JOBCTL_TOKEN (can be created at app.jobctl.app/settings).
Privacy and security:
- Data is stored only in the user's own account; full JSON export and GDPR account deletion are built in.
- The API is deterministic, and the server does not make LLM calls; the AI layer is handled by the user's own agent.
- Every change is an auditable event and is tagged with the actor (agent / human / system).
Use cases: suitable for developers who are job hunting and need to track multiple applications and interview progress at the same time, especially users who already use Claude Code or other MCP clients and want to maintain their job search records in natural language.
Limitations: the materials do not provide the specific list of tools, the parameter details of each tool, whether it is free or paid, or platform compatibility and other further information, so these are not yet provided.