Who Knows Whatby ASU Prep

find_alignments

Use find_alignments to start from one source standard and retrieve likely matches in one or more target frameworks. Results include the target standard, alignment type, confidence, and an LLM-generated rationale for why the match holds. A common use case is mapping a district science standard to CCSS literacy and math supports before building a crosswalk or lesson plan.

Parameters

  • Name
    source_standard_id
    Type
    string
    Description
    The canonical WKW standard ID for the source standard. Use this when you already have the exact standard record; do not combine it with weaker identifiers unless you need disambiguation safeguards.
  • Name
    source_csp_id
    Type
    string
    Description
    The source standard's CSP identifier, if available. This is useful for integrations that store standards by provider or namespace IDs rather than WKW standard IDs.
  • Name
    source_human_coding_scheme
    Type
    string
    Description
    The human-readable code for the source standard, such as "MS-PS2-2" or "6.RP.A.2". Pair it with source_framework_guid when the code may exist in more than one framework.
  • Name
    source_framework_guid
    Type
    string
    Description
    The framework GUID containing the source standard. Required when resolving a source_human_coding_scheme that is not globally unique, and useful as a constraint for source_csp_id lookup.
  • Name
    target_framework_guids
    Type
    string[]
    Description
    A list of framework GUIDs to search for aligned target standards. Omit to search across all available target frameworks, but prefer an explicit list to reduce noise and latency.
  • Name
    alignment_types
    Type
    string[]
    Description
    The alignment evidence types to include. Defaults to ["embedding+llm","human_confirmed"]; use "human_confirmed" when you only want reviewed mappings, or include "embedding+llm" for generated candidates with rationale.
  • Name
    min_confidence
    Type
    number
    Description
    Minimum confidence score for returned alignments, from 0 to 1. Defaults to 0.6; raise it for review workflows and lower it when exploring possible crosswalks.
  • Name
    limit
    Type
    number
    Description
    Maximum number of alignments to return. Defaults to 20; use a smaller value for agent turns and a larger value for batch review screens.

Example

{"source_human_coding_scheme":"NGSS.MS-PS2-2","source_framework_guid":"fw-ngss-2013","target_framework_guids":["fw-ccss-math-2010","fw-ccss-ela-literacy-2010"],"alignment_types":["embedding+llm","human_confirmed"],"min_confidence":0.72,"limit":5}

Common pitfalls

  • Do not rely on source_human_coding_scheme alone when codes are reused across frameworks or editions; include source_framework_guid.
  • A high confidence score is not the same as full equivalence. Check the relation and rationale before treating an alignment as substitutable.
  • Leaving target_framework_guids empty can return broad matches from unrelated frameworks; constrain targets when the agent has a known destination.