TEA: Test Engineering Architect
Documentation · Getting started · Contributing · MIT license
TEA is a BMad module for test planning, automation, evaluation, and release decisions.
Its agent, Murat, uses eight skills and a shared testing knowledge base to turn requirements and risks into tests and evidence.
You can use TEA on its own or alongside BMad Method.
TEA helps you:
- Plan coverage by risk and priority, from P0 to P3.
- Set up a test framework and CI, write acceptance tests, and expand automation.
- Review test quality against a fixed criteria registry.
- Evaluate skills, agents, and AI features with scored probes and regression checks.
- Trace requirements to evidence and make PASS, CONCERNS, or FAIL release decisions, with waivers recorded separately.
Install
Install the skills from GitHub:
npx skills add bmad-code-org/bmad-method-test-architecture-enterprise
Setup needs uv and the bmad and bmod-core-tools skills from BMad Method core.
If those skills are missing, install them:
npx skills add bmad-code-org/BMAD-METHOD --skill bmad bmod-core-tools
Then type this in your assistant chat:
bmad setup tea
Setup asks about your test stack, integrations, and output folder, then writes the answers to _bmad/config.toml.
Run it again to change those answers.
The configuration reference covers team and personal overrides.
The GitHub install follows main.
To install a tagged release, replace <version> with the release number:
npx skills add https://github.com/bmad-code-org/bmad-method-test-architecture-enterprise/tree/v<version>/skills
Upgrading to 3.0
The eight direct workflow commands now share the bmad-tea-* prefix. Replace bmad-testarch-automate, -evaluate, -framework, -nfr, -test-design, -test-review, and -trace with the same suffix under bmad-tea-. Replace bmad-teach-me-testing with bmad-tea-teach-me-testing. The 3.0 package does not publish the old slash commands. The bmad-tea agent, bmod-tea module, menu codes, and tea-* CLI executables keep their names.
After installing 3.0, move project customizations to the new names:
npx --package=bmad-method-test-architecture-enterprise@3 tea-migrate-customizations --project-root . --check
npx --package=bmad-method-test-architecture-enterprise@3 tea-migrate-customizations --project-root .
This includes old CI and ATDD customization files and old skill references in custom bmad-tea agent menus. The command moves old-only files with their comments and permissions intact. When both old and new filenames exist, it reports both paths and asks for a manual merge so no comments or values are lost. Resolve the reported files, then rerun the command. See the configuration reference for details.
An existing host may retain the old skill folders after installing 3.0. Check the host's project and global skills folders, back up any edits in the old folders, and move those eight folders out of skill discovery. The troubleshooting guide lists the exact folders and a safe cleanup sequence. Restart the host so its command list refreshes.
Start with one skill
Eight commands start the eight workflows: Framework includes CI setup; Automate includes red and expand modes.
In Claude Code, Cursor, and Windsurf, type /bmad-tea-test-design in chat.
In Codex, type $bmad-tea-test-design.
You can also load /bmad-tea or $bmad-tea and choose a menu code.
Each skill can run directly in a fresh session.
| Skill | Command | Menu | Use it to |
|---|
| Automate | /bmad-tea-automate | TA | Generate red acceptance scaffolds or expand coverage |
| Test Review | /bmad-tea-test-review | RV | Audit test quality and score findings |
| Test Design | /bmad-tea-test-design | TD | Plan risks, coverage, and NFR evidence |
| Framework | /bmad-tea-framework | TF | Set up a test framework, CI, or both |
| NFR | /bmad-tea-nfr | NR | Audit implemented NFR evidence |
| Trace | /bmad-tea-trace | TR | Map requirements to tests and decide a release gate |
| Teach Me Testing | /bmad-tea-teach-me-testing | TMT | Learn testing through seven sessions |
Evaluate:
- Evaluate:
/bmad-tea-evaluate, menu EV. Build and run a behavioral evaluation.
Framework and CI setup share one skill, bmad-tea-framework.
Ask for framework only, CI only, or both; TEA infers the scope from your prompt and asks "Do you want CI too?" once when CI scope is unclear in an interactive session. An unattended request with unclear scope runs framework setup only and states that CI was excluded.
Request CI setup through /bmad-tea-framework or the agent's CI menu code. Migrated CI customizations remain usable.
Create, Resume, Validate, and Edit remain available for the selected setup scope.
Automation uses red mode before implementation and expand mode for existing code. Request red mode through /bmad-tea-automate or the agent's AT menu code; the direct command defaults to expand. Migrated customization files and saved progress remain usable. Create runs execute the generated tests and repair test issues for up to three rounds. Red confirms the intended missing behavior; expand aims for passing coverage and reports product defects.
For Codex, replace the leading / in the table with $.
The agent menu also accepts GATE to route you through test review, NFR evidence audit, and trace Phase 2.
Choose a starting path:
How it works
Each skill loads one step file at a time.
The steps define the task, the evidence to collect, and when to stop.
They load the knowledge fragments needed for that task from skills/bmod-tea/knowledge/tea-index.csv.
Skills can delegate independent tasks when the assistant supports workers; execution mode is configurable.
Review findings use fixed criteria and severity rules.
Trace applies coverage thresholds and evidence checks to the release decision.
Generated tests and reports still need review, and a release gate depends on the evidence collected.
See step-file architecture, test quality standards, and risk-based testing.
Planning and traceability work across stacks.
Test generation and execution support vary by target.
The execution-target matrix lists the supported stacks and their limits.
Command-line tools
TEA also publishes an npm package for CI and terminal use.
It requires Node.js 22.20.0 or later:
npm install --save-dev bmad-method-test-architecture-enterprise
tea-teach teaches one learner turn and keeps conversation and session progress.tea-test-review runs the review skill against changed tests and returns a gate verdict.tea-framework creates, resumes, validates, and edits test framework and CI setup with retained run evidence.tea-trace maps requirements to existing evidence and returns a coverage gate.tea-evaluate checks, runs, scores, and compares behavioral evaluations.
For tea-evaluate, install eval-quality alongside TEA in the evaluations folder:
npm install --prefix evals bmad-method-test-architecture-enterprise eval-quality
npm exec --prefix evals -- tea-evaluate --help
The npm package includes the skills used by the tools.
See the CI examples for test-review integration.
Contributing
CONTRIBUTING.md covers local checks, pull requests, and publishing.
How TEA Is Tested explains the test and evaluation layers.
The test suite guide has commands for running individual suites and live evaluations.
Community
You can support BMad by starring the repository, buying a coffee, or contacting [email protected] about sponsorship.
License
MIT.
