feat: interactive autograded starter workspace (Module 10 — AI Software Engineering)

[IshmaelRogers]

Summary

Converts Module 10 (AI Software Engineering) from LMS lesson content into a self-contained, interactive autograded starter workspace, mirroring the approved Module 04 template (vitest + strict TypeScript).

Students clone, read each exercise's README.md, implement the // TODOs in its src/, run npm test, and submit. The tests are the spec — red → green — and no answer keys are shipped.

What changed

• Removed all LMS markdown (Lessons, learner/instructor guides, syllabus, rubric, dashboard) and the flat lab markdown + labs/aitools.mjs (which contained full working reference implementations). Kept resources/, .devcontainer/, and the assignments/*submission-template*.md files.
• Toolchain copied from Module 04: package.json, tsconfig.json, vitest.config.ts, scripts/grade.mjs, .github/workflows/autograde.yml, .gitignore. The grader's strict type gate now uses the root tsconfig.json only (no Module 04 lab-06 strict pair).
• 8 lab folders + capstone, each self-contained (README.md, src/ stubs with // TODO, tests/ executable spec):

Exercise	You implement
labs/lab-00-setup	formatCents
labs/lab-01-context-engineering	contextComplete
labs/lab-02-architecture-prompting	checkArchitecture (fitness function)
labs/lab-03-ai-assisted-tdd	slugify (test-first)
labs/lab-04-code-review	applyDiscount (find & fix the planted bug)
labs/lab-05-refactoring	total (characterization)
labs/lab-06-orchestration	runWorkflow / actionInScope
labs/lab-07-security-governance	scanForSecrets / detectInjection / governanceGate
assignments/capstone	reviewChange — integrates context + architecture + governance

The capstone's reviewChange returns { approved, reasons } (approved only when context is complete and architecture conformant and governance passes), with a *.test-d.ts for the result type.

Verification

• Stub state: node scripts/grade.mjs → 12/39 tests (31%), exit 1 (RED). Strict type-check clean. (The 12 passing are degenerate/empty-input cases plus the capstone type assertions; every exercise shows ❌.)
• Reference solution (implemented in a throwaway temp dir, then deleted): 39/39 tests (100%), exit 0, strict type-check clean.

Notes

• All src/*.ts are strict-TS-compilable stubs returning wrong/empty values — no any, no as, no @ts-ignore.
• Lab 04's stub deliberately embodies the naive buggy price - price*pct so the review tests fail until clamped.
• Per-lesson and capstone written reports are still submitted via the LMS using the assignments/*submission-template*.md files (not autograded here).

[github-actions]

Forge SWEXP — Module 10 autograde

Score: 12/39 tests (31%) · Strict type-check: ✅ clean

Exercise	Tests	Status
assignments/capstone	3/9	❌
labs/lab-00-setup	0/3	❌
labs/lab-01-context-engineering	1/4	❌
labs/lab-02-architecture-prompting	2/3	❌
labs/lab-03-ai-assisted-tdd	0/2	❌
labs/lab-04-code-review	1/3	❌
labs/lab-05-refactoring	1/3	❌
labs/lab-06-orchestration	1/4	❌
labs/lab-07-security-governance	3/8	❌

Keep going — open each exercise folder, implement the // TODOs in its src/, and run npm test. The tests in each tests/ folder are the spec.