Securing an agent that holds a signed-in LinkedIn session and my resume: threat model, Keycloak, OpenBao, a read-only tool allow-list and deterministic claim checks.

One FastAPI process with six agents, a LinkedIn browser in a sidecar, a local model behind LiteLLM and Supabase, running in one pod on the homelab platform.

A three-tab review UI in plain HTML and JavaScript, designed around one evening habit: read the queue, check the claims, approve or reject.

How Job Hunter Agent went from three articles to a live homelab app in two days, and the phases and steps that came after.

What Job Hunter Agent must do and how well: discovery, scoring, truthful tailoring and a manual approval gate, with the rules that keep it honest.

Why I built a job-hunting agent that does the searching and drafting but never applies on its own, and how I'll know it's worth running.

Every architecture decision behind Job Hunter Agent, with its context, the options considered and the consequences.

Warehouse, lake, lakehouse, fabric and mesh compared by data flow, ownership, governance and consumption, with guidance on when to pick each.

How payers review whether care is medically necessary (prospective, concurrent and retrospective review) and why it leads to denials.

When pre-approval is required, how a request moves between provider and payer, and why delays and denials happen.