For subscribers
Your feed, inside your own tools.
JobFeeder speaks the Model Context Protocol, so Claude, Cursor or anything else that supports it can read the roles we scored for you — with the reasoning attached. Ask your agent what came in this morning and have it draft the outreach, without leaving the thing you already work in.
It is read-only on purpose. An agent can read what you were sent and why; it cannot change what you are looking for. That stays in the app, where you can see it.
Connect it
Create an access token in the app under Settings, then point your client at the server. In Claude Code that is one command:
claude mcp add --transport http jobfeeder \
https://jobfeeder.app/api/mcp \
--header "Authorization: Bearer jfm_your_token"Any other MCP client wants the same two things: the URL https://jobfeeder.app/api/mcp over HTTP, and your token in an Authorization: Bearer header.
Tokens are shown once. Rotating one disconnects every agent using the old one; revoking takes effect immediately.
What it exposes
- today_matches
- The roles that cleared your bar today, highest score first, each with why it fits and the one concern worth knowing.
- match_detail
- Everything known about one role: full reasoning, concerns, what to emphasise, a suggested next step, and a company snapshot.
- brief_history
- Your past daily briefs — what was sent and when.
- my_profile
- The rubric the screener judges roles against: tracks and title seeds, comp floor, location rules, dealbreakers.
Note what is not here: a search over the job market. JobFeeder reads around 63,000 postings every weekday morning and keeps only what clears your profile, so there is no index to hand over. What is worth having is the judgement — which roles cleared your bar, and why.
It comes with the full feed.
$19.99 a month. Cancel anytime.