standards-match

Returns the 1–5 most relevant Florida standards for a given topic/grade/subject. No generation — lookup and return only. Used by lesson-planner, assessment-designer, and any skill needing verified standard codes before generating content.

Input

{
  "grade":   "4",
  "subject": "Math",
  "topic":   "fractions on a number line",
  "max":     3
}

grade: K–12 or grade band (e.g. "6-8"). subject: Math, ELA, Science, Social Studies, ELD, etc. topic: keyword phrase or short description. max: how many standards to return (default 3).

Output

{
  "tool": "standards-match",
  "query": {"grade": "4", "subject": "Math", "topic": "fractions on a number line"},
  "standards": [
    {"code": "MA.4.FR.1.1", "description": "Model and express a fraction including mixed numbers...", "source": "FL B.E.S.T.", "confidence": "high"},
    {"code": "MA.4.NSO.1.1", "description": "Express how the value of a digit changes...", "source": "FL B.E.S.T.", "confidence": "medium"}
  ],
  "match_method": "keyword | L1_cache | uncertain",
  "human_review_required": true,
  "note": "Verify codes against CPALMS before citing in any artifact."
}

match_method is unified_index when the unified offline index is built (python3 tools/offline_index.py --standards "<topic>" --grade <g> --subject <s> — a deterministic zero-token SQLite/FTS5 lookup that returns only the matching rows; see canonical-sources/index/README.md), L1_cache for the standards-only cache (shared/cache/cache.py), keyword for tools/fl_lookup.py direct search, uncertain when no index is built and the match is model-inferred. Prefer the index — it returns verbatim canonical rows (~100–350 tokens) instead of loading the ~470k-token standards corpus or recalling from memory, so it cannot fabricate a code. Never fabricate a standard code — return empty standards and explain in note instead.

How to use this atom

"Find the top 3 Florida B.E.S.T. Math standards for grade 4 fractions on a number line."
→ standards-match returns the JSON above
→ caller passes standards[] to objective-write or activity-generate

Do NOT use this atom for

• Generating any instructional content (use objective-write, activity-generate)
• NGSSS or Common Core lookup (label the framework in subject if non-BEST)
• Confirming whether a standard was repealed — route to standards-updater

Pipeline note

Output conforms to references/metadata-schema.md. Follows references/method.md at the Standards step only. No Generation or Differentiation step — this atom stops at Standards Alignment and returns. human_review_required: true because model-inferred matches are not guaranteed correct.