QE-MCP
A strategy-driven Model Context Protocol server that gives coding agents repository-specific rules for analysing test gaps, planning tests, validating generated code and enforcing quality gates.
Quality workflow
Core capabilities
Repository analysis
Pairs source and test files, identifies gaps and prepares structured context for deeper agent analysis.
Prescriptive test planning
Creates explicit test cases and behavioral specifications rather than asking the model to invent an unconstrained suite.
Generation briefs
Assembles strategy-specific instructions that tell an agent how tests should fit the repository's expected architecture and conventions.
Layered validation
Checks generated tests for syntax, naming, specification completeness, assertions, lane compliance, unsafe bare assertions and skip limits.
Hard enforcement
Combines analysis and validation into an acceptance gate that can fail when required coverage or architectural rules are not met.
Change-focused analysis
Includes dependency and diff-oriented analysis capabilities so enforcement can be focused on changed code where appropriate.
Strategy-based instead of prompt-only
QE-MCP separates quality policy from an individual model prompt. Built-in strategies include Python/pytest rules and an Android Compose strategy with distinct test lanes and source classification. Custom strategies can extend the same mechanism for other repositories.
The intent is not to make an agent autonomous by default. It is to make the evidence and rules around generated tests explicit enough that an agent's output can be checked before it is accepted.