FairHire is an open-source applicant tracking system built on one idea: a paid work sample should outweigh the interview. Weighted scorecards are a first-class object, every decision publishes its fairness trail, and an AI agent can run the whole pipeline through MCP.
The paid trial is a first-class object: scope, deliverables, hours, rate, and a rubric score on the actual work shipped. It carries the most weight in the composite, and you pay candidates directly. FairHire never takes a cut.
Define criteria and weights up front. Every score contributes score × weight to a transparent composite. The same rubric, applied equally to every candidate.
Login-gated transparency pages snapshot the scores, weights, and stage history behind each decision at publish time. Defensible by design, for candidates and committees.
FairHire exposes a first-class fairhire.* MCP tool catalog over the same guarded service layer the UI calls — two callers, one service. Your agent runs the funnel as a real actor, while consequential actions like offers and approvals stay human-gated.
FairHire is MIT-licensed and open source: the composite math, scorecard weighting, and hash-chained audit trail are all readable, and you can self-host and own your data end to end. The public repository is on its way.
Run distinct, org-scoped workspaces from a single deployment. Every candidate, note, and decision is isolated per company — no cross-tenant leakage, by construction.
Define the criteria that matter and how much each one weighs before the first candidate applies. The rubric is set once and applied equally.
Move candidates through source → screen → work-sample → paid-trial. The paid trial is real, compensated work and carries the most weight in the composite.
The composite math is transparent. Extend an offer and publish a login-gated fairness page that snapshots exactly how the decision was reached.
FairHire ships a 200-tool fairhire.* MCP catalog — two
callers over the one guarded service layer the UI calls — so an AI agent
can source roles, move candidates, score work samples, and read the
pipeline. Everything a
recruiter can do in the UI, over the same permission-checked, audited
surface. Offers, rejections, and approvals stay human-gated, so
automation never crosses the lines that matter.
> fairhire.req.list({ status: "open" })
> fairhire.stage.advance({ candidateId, toStage: "judgment" })
> fairhire.trial.score({ trialId, score })
> fairhire.transparency.assemble({ jobId })
# advance → offer · reject → human-gated ✋Open source, transparent, and ready for your team and your agents.