Available now live
JobsRadar
Every company runs its own job board on its own applicant tracking system. JobsRadar asks all of them at once and hands back one deduplicated, filtered list.
The problem it removes
One call per job, not one call per website.
A recruiting or job-search agent otherwise has to know that this company uses Greenhouse, that one uses Lever, and that a third moved to Ashby last quarter, then reconcile four different field names and drop the duplicates itself.
- Every board, one shape.Greenhouse, Lever, Ashby and Workable normalize to the same posting: title, company, location, remote flag, department, publication date, apply URL. The hosted endpoint serves Greenhouse, Lever and Ashby, for any board slug you supply and not only the tracked ones. Workable is implemented and works in the open-source build, but it answers HTTP 429 to every request from Cloudflare's network, which is where this endpoint runs, so we do not pretend otherwise.
- Deduplicated. The same role published on two boards, or twice on one, collapses to a single entry.
- Honest coverage. The response says how many boards answered and names any that failed, so a broken board never silently shrinks your results.
- Dated. Tracked boards are refreshed every 15 minutes, so a search reads one copy instead of waiting on every board in turn. Each answer carries
data_as_of, the time its oldest board was fetched; a board whose copy is more than an hour old is fetched live instead. - A curated directory. 68 companies, each resolved against the live board API, so you do not need to know which applicant tracking system any of them uses. A daily job walks every entry and fails when one goes stale, because companies migrate and a slug that quietly stops answering is worse than one that is missing.
Tools
Three tools, plus free usage.
| Tool | Credits | What it does |
|---|---|---|
| search_jobs | 1 | Search open roles across every tracked company board in one call, deduplicated, with filters on title keywords, location, remote and publication date. Boards refresh every 15 minutes and each answer says how old it is. Public board APIs. Greenhouse, Lever and Ashby, for the curated directory and for any board slug you supply. |
| get_company_jobs | 1 | List every public posting at one company. Give a tracked name or any board slug and the applicant tracking system is detected for you. Greenhouse, Lever or Ashby, detected from the slug. |
| list_supported_companies | 0.2 | Read the curated directory of tracked company boards and the applicant tracking system behind each, so you know what search covers. Our directory, re-checked against the live APIs. |
| get_usage | free | Read the remaining credit balance and the last 30 days of calls. Your account. |
Measured
100% of scenarios passed on the last run.
11 agent scenarios against the deployed server, failure cases included. p95 1.34s end to end, which includes the job boards themselves.
Connect
The endpoint.
A remote MCP server over Streamable HTTP. Any MCP client reaches it, and so does plain HTTP if you would rather call it directly. Swap in the key from your free account and the first call works.
https://jobsradar.agenttoolworks.com/mcp
claude mcp add --transport http jobsradar \ https://jobsradar.agenttoolworks.com/mcp \ --header "Authorization: Bearer atw_live_your_key"
Pricing
Same credits, same balance.
Credits are shared with every other server, so one balance covers everything you call. Packs start at $10 for 1,000 credits and never expire, and 100 credits are free without a card.
Where the data comes from
Only the official public job board APIs that applicant tracking systems publish so employers can distribute their openings: Greenhouse, Lever and Ashby on this endpoint. No login, no anti-bot circumvention, no candidate data, and every result links to the employer's own apply page.
We do not touch LinkedIn, Indeed or any source behind a login. That line is in the spec and this product stays well inside it.