feat(ts-sdk): jurisdiction support for math standards alignment

[adnanrhussain]

Summary

• Adds Jurisdiction enum (52 values, matching KG API) to the KG client layer so future evaluators can reuse it
• Makes jurisdiction a mandatory top-level parameter on all evaluator entry points (evaluate, evaluateItems, evaluateQuestionBank, evaluateByGrade) — forces explicit choice, no silent Multi-State default
• KG client is now subject-agnostic: academicSubject is passed by the caller; math evaluator passes 'Mathematics' internally via KG_SUBJECT
• Switches from throw-on-ambiguous to limit=1 + take-first for standard lookups — eliminates deduplication logic and matches API design intent
• Removes grade from the public evaluate() signature and from system/user prompt templates; grade is still recorded internally when available (via evaluateItems/evaluateQuestionBank) but is grade?: string on StandardAlignmentResult
• Framework UUID lookups for non-Multi-State jurisdictions use a cached call to /standards-frameworks

Test plan

• npm test — 278 pass, 0 fail
• npm run lint — 0 errors
• evaluate(question, statementCode, Jurisdiction.California) passes jurisdiction=California and academicSubject=Mathematics to the KG search endpoint
• evaluateByGrade(questions, '3', Jurisdiction.California) calls /standards-frameworks?jurisdiction=California&academicSubject=Mathematics to get the correct framework UUID

[Copilot]

Pull request overview

Adds jurisdiction-aware standards alignment to the TypeScript math evaluator and Knowledge Graph client so callers must explicitly choose which state/adopted framework to use (instead of silently defaulting to Multi-State/CCSS).

Changes:

• Introduces a shared Jurisdiction enum in the KG client layer and re-exports it for SDK consumers.
• Updates math standards alignment evaluator entry points to require jurisdiction, removes grade from the single-question evaluate() public signature, and threads academicSubject into KG calls.
• Adjusts KG standard lookup behavior to use limit=1 and take the first match; adds framework UUID lookup/caching for non–Multi-State jurisdictions.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
sdks/typescript/tests/unit/knowledge-graph/client.test.ts	Updates KG client behavior expectations for multi-result standard lookups.
sdks/typescript/tests/unit/evaluators/math/standards-alignment.test.ts	Updates evaluator tests for new signatures and verifies jurisdiction/subject are passed to KG.
sdks/typescript/src/prompts/math/standards-alignment/index.ts	Removes grade as a system-prompt placeholder input.
sdks/typescript/src/knowledge-graph/types.ts	Adds Jurisdiction enum (string values matching KG API).
sdks/typescript/src/knowledge-graph/index.ts	Re-exports Jurisdiction and option types from KG client.
sdks/typescript/src/knowledge-graph/client.ts	Adds jurisdiction/subject-aware options, standard lookup limit=1 behavior, and framework UUID caching for standards-by-grade.
sdks/typescript/src/evaluators/math/standards-alignment.ts	Makes jurisdiction required across public entry points, removes grade from evaluate(), and passes subject/jurisdiction to KG.
evals/prompts/math/standards-alignment/user.txt	Removes grade from the user prompt template.
evals/prompts/math/standards-alignment/system.txt	Removes grade placeholder from system prompt template.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov]

Codecov Report

❌ Patch coverage is 84.96732% with 23 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
sdks/typescript/src/knowledge-graph/client.ts	62.50%	15 Missing ⚠️
...escript/src/evaluators/math/standards-alignment.ts	86.44%	8 Missing ⚠️

📢 Thoughts on this report? Let us know!

[adnanrhussain]

Making this consistent back w/ the source notebook

[adnanrhussain]

Making this consistent back w/ the source notebook. Minimal changes enable evaluating multiple learning components, vs a single component

[czi-fsisenda]

LGTM! 🚀

[czi-fsisenda]

[P2] Been trying to establish a standard output shape for evaluation results. Could it work for this?

evaluators/sdks/python/src/learning_commons_evaluators/schemas/evaluator.py

Lines 194 to 220 in 70362e4

	class EvaluationAnswer(BaseModel):
	"""The main answer of an evaluation: score and label."""
	score: Any = Field(
	description="The score of the evaluation. This is typically a string or a number."
	)
	label: str = Field(
	description="The label of the evaluation. This is typically a human-friendly string."
	)
	class EvaluationExplanation(BaseModel):
	"""Explanation of the evaluation: summary (markdown) and optional keyed details."""
	summary: str = Field(description="A summary of the evaluation in markdown format.")
	details: dict[str, Any] = Field(
	default_factory=dict,
	description="Optional keyed details of the evaluation.",
	)
	class EvaluationResult(BaseModel):
	"""Standard evaluation result: answer, explanation, and metadata."""
	answer: EvaluationAnswer
	explanation: EvaluationExplanation
	metadata: EvaluationMetadata

[czi-fsisenda]

Generated, right?

[adnanrhussain]

the client is not generated, since I have some caching, concurrency limits, etc. But I could do a followup and break that apart, and just have the generated faw fetch and the wrappers on top