Wispr Flow: Dictation That Works EverywhereStop typing. Start speaking. 4x faster.
Promoted
AI Applyd MCP
Your job search inside Claude, with verified auto-apply
9 followers
Your job search inside Claude, with verified auto-apply
9 followers
A remote MCP server that puts ten job-hunt tools inside Claude, Cursor, and ChatGPT: ATS resume scoring, job-description analysis, interview prep, cover letters, job search, and auto-apply. The auto_apply tool runs a pipeline that verifies each submission actually registered on the employer's hiring system and returns an honest errors count. One URL, OAuth 2.1, sign in with Google once. Resume scoring and job analysis never consume AI credits.
one honest note, a few hours in, because it is the thing most likely to trip someone up.
auto_apply is a kickoff, not a synchronous call. it starts the run and hands back counts while the verification finishes in a pipeline behind it. so if you call it and read the numbers straight away, you are reading an early state, not a final one. the dashboard settles a minute or two later. i have not found a good way to say "still settling" inside a tool result that assistants render sensibly, and if anyone here has solved that, i want to hear it.
the two tools that cost nothing are the ones worth trying first. score a resume against a real posting, or ask what a job description is actually asking for. neither spends ai credits, so you can judge whether the output is any good before deciding anything.
and the errors count is not decoration. applications do fail, and it will tell you when one did.
The honest errors count is a really nice touch, that alone would make me trust this over the usual "we applied to 300 jobs" black box. One thing I'd love: let me set per-application guardrails like minimum salary, max commute, or visa sponsorship required, and have auto-apply skip anything that doesn't match before the ATS submission happens. Saves me from having to babysit the pipeline.
@compliancepro that is already how it works, and the fields turned out closer to your list than i expected when i read it.
the preference set gates matching, which sits upstream of submission, so anything failing it never reaches the ats at all. salary is a real min and max with a currency, normalised to usd so postings in different currencies compare properly. commute is a location plus a search radius in km rather than a time. visa sponsorship is an explicit four-way: any, not needed, preferred, required.
the honest caveat, because you would hit it: those gate what gets matched and queued, not a second hard re-check at submit time. so if a posting's salary is simply absent rather than low, it can still get through, because we cannot filter on a field the employer never published. that is the real gap in what you are describing.
the skipped count in the run result is where filtered ones land, kept separate from errors so you can tell "we chose not to" from "it failed".