Quality Supervisor

QA skills for your coding agent

Your agent can already reach Qase through MCP. Quality Supervisor gives it the analysis a QA lead would run on that data.

Start trialTalk to sales

No credit card required

/plugin marketplace add qase-tms/qase-quality-supervisor
ISO/IEC 27001·SOC 2·GDPR·Enterprise SLA

What your AI harness needs to protect quality

Connecting an agent to your test data is the easy part. Making it useful, safe and checkable takes three more things.

Knowing what to ask

Access to runs, results and defects doesn't tell an agent how to find a coverage gap or call a release. Without packaged analysis, every team writes those prompts from scratch.

Guardrails in code

A prompt that says “don't delete anything” is a suggestion. Limits on what an agent can do with your test data belong in the tooling, not the instructions.

Answers you can check

A finding without the query behind it is a guess you can't audit. You need to see what was asked and be able to run it yourself.

Six skills, one plugin

Quality Supervisor is an open-source plugin for MCP-capable AI clients. Each skill packages one piece of QA analysis on top of your Qase data.

Coverage gaps

“Where are our coverage gaps?” Finds requirements, suites and critical paths with no tests, then drafts the missing cases for you to approve.

Failure triage

“Triage the latest run and tell me what's a real bug.” Groups failures and sorts each one into product bug, automation issue, environment problem or flake. Creates and links defects when you say so.

Flaky tests

“Which tests should we quarantine?” Measures instability from result history, confirms with re-runs, and recommends quarantine or a fix for each test.

Release readiness

“Are we ready to ship milestone 2.3?” A go/no-go check across five dimensions for a milestone, test plan or release.

Quality pulse

“Give me a quality pulse for the last two weeks.” An HTML report card covering execution, pass rates, defect backlog and test case activity.

Change impact

“What should QA test on this branch?” Reads the diff, traces the affected code paths, and returns a regression checklist plus the new tests the change needs.

Want all of it at once? Run with your project code for coverage, flakiness, triage and readiness in one read-only pass.

Guardrails in the plugin, not the prompt

Quality Supervisor reads by default and asks before it writes.

No delete operations

A hook blocks every Qase delete tool before it runs, on macOS, Linux and Windows. That holds no matter what the prompt says.

Writes wait for you

Analysis is read-only. Drafting cases, creating defects or linking them to results happens only after you confirm.

The query behind every finding

Each finding comes with the QQL query that produced it, so you can run it in Qase and check the same data yourself.

Setup

Up and running in three steps

Quality Supervisor runs in Claude Code, Claude Desktop, Cowork, OpenAI Codex, Cursor, VS Code and Atlassian Rovo.

  1. 1

    Connect the Qase MCP server

    Quality Supervisor works through the Qase MCP server. Set up MCP first if your client isn't connected yet.

  2. 2

    Install the plugin

    Steps for Codex, Claude Desktop, Cowork and other clients are in the help center. In Claude Code, add the marketplace and install the plugin:

  3. 3

    Ask in plain language

    Try “Triage the latest run” and Quality Supervisor routes it to the right skill. Or call a skill directly:

Ship quality software faster with Qase

Start trialTalk to sales

FAQs

An open-source plugin that adds QA analysis skills to your AI client. It works on top of the Qase MCP server: MCP gives your agent access to Qase data, and Quality Supervisor packages the analysis a QA lead would run on it, from coverage gaps to release readiness.
Quality Supervisor is included on Teams and Enterprise. The Free plan has limited access, because the analyses that rely on QQL queries need Teams or Enterprise.

Compare plans →
Claude Code, Claude Desktop, Cowork, OpenAI Codex, Cursor, VS Code and Atlassian Rovo, plus other MCP-capable clients. It connects to Qase through the Qase MCP server.
Connect the Qase MCP server first. Then add the qase-tms/qase-quality-supervisor plugin marketplace in your client and install the quality-supervisor plugin. The help center has the steps for each client.

Install guide →
Only one skill does. Change impact reads your branch's diff to work out what QA should test, and you decide when to run it. The other five skills work from your Qase data alone.
Every change made through MCP runs under your Qase permissions and shows up in the Qase UI (and in audit logs on Teams and Enterprise), same as a change made by hand. Quality Supervisor goes further: it has no delete operations, and anything it writes waits for your confirmation.
Ask your assistant which query it ran. Quality Supervisor includes the QQL query behind every finding, so you can run it in Qase and check the same data yourself.
Your test data stays in Qase, which remains the system of record. When you ask a question, the results your assistant pulls are sent to the AI model your client uses, under that provider's terms. Qase is ISO/IEC 27001 and SOC 2 certified and GDPR compliant.

Trust center →
Yes. It's MIT-licensed, and the source for every skill is on GitHub.

View on GitHub →