How the Bryq MCP Works: Assessment Data in Your AI Assistant

What the Bryq MCP actually does, and how it works

Bryq's MCP is live and included in every Bryq plan. It connects your Bryq account to the assistant your team already works in. This is the longer explanation: what the connection reads, what it deliberately cannot do, how access is scoped, and what becomes possible when assessment results sit next to your own hiring documents.

What the connection cannot do

The connection is read-only. Here is what that rules out.

It reads from your Bryq account and returns what it finds. It does not create an assessment. It does not send an invitation. It does not move a candidate to another stage. It does not write a note, change a score, or alter a single record in Bryq.

Read-only is the current scope, and everything here describes what the connection does today. A connection that only reads is a much smaller thing to review, because the worst case is that somebody sees data their own Bryq account already permitted them to see.

How access actually works

The connection is added inside the assistant, not inside Bryq. Someone on your team opens Claude, ChatGPT, Copilot, or any assistant that supports the Model Context Protocol, adds the Bryq connection, and signs in with their Bryq account.

That sign-in is the whole permission model. Bryq already has six default roles, including Administrator, HR Administrator, Hiring Manager and HR User, plus any custom roles a company has configured. Through the connection, each person sees what their own Bryq account already permits, and nobody's access widens by connecting an assistant to it. Same data, same permissions, reached from somewhere else.

What the connection reads

Three things, worth separating because they do different work.

1. Your results. The candidate outcomes your Bryq account holds, from Profile Fit and from Skill Tests, including AI Fluency.

2. Bryq's scoring and recommendations. Not the raw numbers alone. The scoring already sitting in the account, and what Bryq recommends on the basis of it.

3. Scope and filters. One candidate, a set of them, or everyone who applied to a role, narrowed by job, department, date, score, candidate name or hiring stage. Date filters bound assessment completion rather than when someone applied.

Three things this makes practical

These are the patterns the connection supports.

Working a whole role in one answer instead of one screen at a time. Pull everyone who applied, or a set of candidates you name, and get it back as a single answer rather than a list view you page through. The work of assembling it is what disappears, not the data, which was always there.

Reading a candidate against the role as it was actually described. This one has a condition attached, so it is worth being plain about it. If the job description, the intake notes or whatever else the role was really built from are open in the same chat, then the assessment result and the original brief can be read together and the places where they pull apart become visible. If those documents live only in your ATS or a notebook, the assistant cannot see them and this does not apply. Bryq does not retrieve any information from your AI assistant.

Asking in your own words rather than picking from a menu. "Give me the top 5 candidates for the Senior Account Manager role, with their strengths and weaknesses" is not a report anyone built in advance. Neither is the follow-up you think of after reading the answer. The connection returns the results and Bryq's scoring; the shape of the question is yours, and it can change every time.