QualityFlow turns issue requirements into human-reviewed test plans and test code

An Aug. 23 commit added a FastAPI dashboard, organization-level rollups and project onboarding to QualityFlow, a Red Hat Community AI Tools framework for moving from a Jira ticket or GitHub issue to a test plan, test description and generated test implementation. The project’s README presents it as community tooling for Claude Code and Cursor AI, not as a supported Red Hat product.

What changed on Aug. 23

The commit introduces team-level pipeline and coverage views plus a manager view that polls peer instances and merges their rollups without a shared datastore. Peers can be configured in YAML or through an environment variable. It also adds a project-onboarding script and template.

That dashboard sits above an existing staged workflow. /stp-builder gathers issue context and produces a software test plan. Separate review and refinement commands evaluate that plan before /std-builder converts it into a YAML test description and test stubs. A final generation step writes the test implementation, while /fix-pr can apply review feedback to planning documents.

Inputs can be Jira identifiers, Jira URLs, GitHub issue URLs or repository-and-issue short forms. The documented orchestration separates collection, regression analysis, plan generation, formatting and review into distinct agents and reusable skills.

The decomposition matters more than the slash commands. It gives quality-engineering teams places to insert project rules and human approval rather than asking one coding agent to infer requirements, test scope and implementation in a single pass.

Where teams configure control

QualityFlow routes Jira prefixes and GitHub repositories to project-specific YAML configuration. Teams can define repositories, components, scope boundaries, test tiers, languages and frameworks. Feature toggles control test-plan generation, test-description generation, optional language-server analysis and personally identifiable information sanitization.

The project also documents explicit review and refinement stages for both the test plan and test description. Its glossary identifies an approval gate where a human must accept a review verdict before the workflow proceeds. That is the practical boundary to inspect before using generated tests in a delivery pipeline.

What to try first

The repository can deploy its resources into Claude Code, Cursor AI or both. It uses MCP servers to retrieve Jira and GitHub context, and can use language servers such as gopls or Pyright to trace code relationships during regression analysis.

A sensible evaluation is narrower than the full pipeline: configure one project, route one issue prefix, generate a test plan and compare its requirement mapping with the team’s existing manual plan. Review the collected issue and repository data, confirm that scope boundaries are honored, and only then enable test-description or code generation.

QualityFlow is best treated as an inspectable community pattern. The Aug. 23 dashboard and onboarding work makes its current development state dateable, but the repository does not establish a product support commitment. Its useful idea is the staged design: AI-assisted quality engineering split into configured steps, visible rollups and explicit review points.