If your team runs hiring on SmartRecruiters, you are working inside one of the broadest ATS surfaces in the market. Sourcing, CRM, career sites, hiring teams, structured interview steps, configurable approval flows, an entire marketplace of partner integrations. That breadth is the reason large hiring orgs pick it. It is also the reason most generic AI integrations land badly. They treat SmartRecruiters as a flat candidate list, ignore the hiring process you actually configured, and write feedback into a free-text comment that no one filters on.

Ribbon MCP is the shortcut. It is a connector that gives any AI agent live, structured access to your SmartRecruiters data and lets the agent write results back without anyone wiring up a separate SmartRecruiters client. Plug it in once and Claude, ChatGPT, Cursor, or whatever agent your team is running can read your pipeline, score candidates, and update SmartRecruiters inside the same conversation.

What MCP is, in one paragraph

MCP stands for Model Context Protocol. It is an open standard, introduced by Anthropic in late 2024, that lets AI models talk to outside systems through one consistent interface, the way USB lets any laptop talk to any printer. The agent does not need to know that your ATS is SmartRecruiters, that your hiring process has six stages with three interview steps each, or that your "Recruiter Screen" is the second step inside the "Interview" stage on a particular job. It asks Ribbon in plain language and gets structured SmartRecruiters data back.

Why this matters specifically for SmartRecruiters

SmartRecruiters has one of the most expressive hiring process models in the category. A job is not just a posting. It carries a hiring team with named roles (recruiter, hiring manager, executive recruiter, coordinator, interviewers), an ordered hiring process with named stages, and inside each stage a set of interview steps with their own scorecards. Source attribution is structured. Approvals route through real people. Career sites are first-class objects with their own audience and branding.

That structure is what an AI agent should be using. When the agent moves a candidate forward, it should know which stage and which interview step come next on the hiring process attached to that job. When it writes a Ribbon interview result back, it should land on the right scorecard against the right step, not in a general comment thread. Ribbon's connector treats jobs, candidates, applications, hiring processes, stages, interview steps, and scorecards as first-class objects, so the agent acts the way a careful recruiter would.

What gets connected

An ATS integration has two halves. What the AI can see, and what it can change. Ribbon covers both, scoped to whatever your SmartRecruiters permissions allow.

On the read side, the agent gets live access to:

• Jobs and career site postings: open and on-hold requisitions, filterable by status, department, location, or hiring team member. The agent sees what is actively being hired for right now, not a stale snapshot.
• Candidates and applications: the full pipeline, with each application carrying its current stage, current interview step, the linked candidate record, source, and any custom properties your team has configured.
• Hiring processes: the stages and interview steps configured on each job, including the order, the scorecards attached to each step, and the rejection reasons your admin has set up.
• Hiring teams: who owns the requisition, who the recruiter is, which interviewers are on the loop, and what role each person plays on the candidate.
• Interview activity and scorecards: what interviews have been scheduled or completed, who the interviewer was, and what feedback has already been written.

On the write side, the agent can:

• Move a candidate forward to the next stage or the next interview step, respecting the order configured on the job.
• Submit a structured scorecard for a specific interview step, with the recommendation, the competency ratings, and the long-form notes mapped to the right fields.
• Reject a candidate with a specific rejection reason from your configured list, plus an internal explanation the recruiter can audit.
• Add internal comments to a candidate or application, visible to the hiring team but not the candidate.
• Update custom properties on a candidate, application, or job where your schema allows.

Every write is attributed, idempotent where it can be, and goes through your existing SmartRecruiters permissions model. The agent is a credentialed actor, not a side-channel.

What this looks like in practice

Five concrete workflows your team can run on day one.

1. Daily pipeline triage. The agent walks every active job, finds candidates sitting in a stage longer than your SLA, and posts a single internal comment summarizing who is stuck, what step they are stuck on, and which interviewer owes feedback. No human builds a pivot table.

2. Async voice screen, results land on the scorecard. Ribbon runs a structured voice interview with a candidate. When it finishes, the agent submits a scorecard to the right interview step with the rating, competency scores, and a transcript link. The recruiter sees it the next time they open the candidate, in the place they already check.

3. "Tell me about my hiring." The hiring manager asks Claude or ChatGPT how their open roles are doing. The agent reads jobs the manager owns, summarizes pipeline depth by stage, calls out roles where time-to-stage is slipping, and offers to draft an internal nudge to the recruiter. No dashboard required.

4. Smart rejection follow-up. When a candidate is rejected with a "skills mismatch" reason, the agent checks whether the candidate fits any other open requisition, drafts an internal note proposing the move, and waits for the recruiter to approve before changing anything in SmartRecruiters.

5. Hiring team prep. The morning of an interview, the agent assembles a one-page brief for the interviewer: the candidate's resume highlights, the previous scorecards on this candidate, the specific step they are interviewing for, and the competencies that step's scorecard expects them to evaluate.

Setup, at a high level

You connect Ribbon to your SmartRecruiters tenant with an OAuth flow scoped to the data your team wants the agent to touch. From there, point any MCP-capable agent at the Ribbon endpoint, log in once, and the agent is operating against your real pipeline. There is no separate connector to maintain, no background sync to babysit, no schema to keep in sync with SmartRecruiters configuration changes. When your admin renames a stage in SmartRecruiters, the agent sees the new name on the next call.

Frequently asked questions

Does Ribbon mirror our SmartRecruiters data anywhere?

No. Reads happen against SmartRecruiters live, scoped to your OAuth permissions. Writes go directly to SmartRecruiters. Ribbon stores its own interview artifacts (transcripts, scorecards Ribbon generated) and references the SmartRecruiters records they belong to, but we do not maintain a shadow copy of your candidate database.

Can the agent only see what the connecting user is allowed to see?

Yes. The connection inherits the SmartRecruiters role of whoever authorized it. If you want a tighter blast radius, create a dedicated SmartRecruiters service user with a scoped role and connect Ribbon as that user. The agent cannot escalate beyond the role you give it.

What about audit trails?

Every write the agent makes shows up in SmartRecruiters with the connecting user as the actor. If you connected as a service user named "Ribbon AI," every move, scorecard, and comment is attributed to that user, in your existing SmartRecruiters audit log, with timestamps. Internally, Ribbon also keeps a log of which agent run made which call.

Does this break our SmartRecruiters Marketplace integrations?

No. Ribbon does not replace anything in your tenant. Your existing sourcing, assessment, background-check, and analytics partners keep running as they are. Ribbon adds an agent-facing interface on top of the same data.

Can we limit what the agent is allowed to write?

Yes. You can scope writes by job, by stage, by action type, or disable writes entirely and run Ribbon as a read-only connector. Many teams start read-only, get comfortable with what the agent is seeing, and turn writes on stage by stage.

Is the agent rate-limited against SmartRecruiters?

Ribbon respects SmartRecruiters rate limits and backs off cleanly when the API pushes back. You will not see the agent burn through your tenant's quota and break unrelated integrations. Bulk operations are batched and paced.

What is next

If your hiring stack already runs on SmartRecruiters, the value is not "add another tool." It is letting any agent your team is already using talk to the pipeline you already have, in the structures you already configured. Ribbon MCP is how that happens. Get in touch to set up a tenant connection, or read the Ribbon MCP documentation to point your agent at it today.