JobOS

· Updated

A local-first, agent-native operating system for job discovery, fit decisions, tailored materials, applications, networking, and follow-up.


JobOS turns a fragmented job search into a durable local system. It brings discovery, evidence-based fit decisions, tailored materials, applications, networking, and follow-up into one terminal interface backed by the user’s own data.

The project is designed around a simple constraint: an agent can help with the work, but it should not silently inherit authority over sensitive career decisions.

What it does

  • Builds a reusable profile from a resume and verified proof points.
  • Imports or discovers roles, then scores fit with explicit evidence and gaps.
  • Drafts role-specific resumes, cover letters, company research, and outreach.
  • Places exact artifact revisions into a human review queue before they become application-ready.
  • Tracks applications, interviews, contacts, tasks, and weekly progress.
  • Connects Hermes, Codex, and Claude Code through shared domain tools and MCP.

Product decisions

JobOS is local-first: SQLite is canonical, readable Markdown and YAML files make the workspace legible to agents, and cloud accounts are not required for the core flow. Human-only approvals remain mediated by trusted terminal input. Credentials, browser sessions, and restricted application answers stay outside agent-readable mirrors.

The interface adapts from a multi-pane dashboard to a focused chat surface and, on compact terminals, separates the dashboard and conversation rather than compressing both into an unusable layout.

Current status

The deterministic workflow, terminal interface, review gates, ACP/MCP paths, and local mirrors are implemented. The first packaged release is being prepared for public installation.