Discover frameworks

Framework discovery is the first step for any source-grounded workflow. Before searching standards or comparing curriculum evidence, identify the accepted framework family, immutable snapshot, graph package, local grade labels, and package capabilities that the request should use.

The primary tools are list_frameworks, get_framework, get_capabilities, and get_framework_statistics.

Recommended workflow

%%{init: {"themeVariables": {"fontSize": "18px"}}}%%
flowchart LR
    A[list_frameworks] --> B[Choose framework and snapshot]
    B --> C[get_framework]
    C --> D[get_capabilities]
    D --> E[get_framework_statistics]
    D --> F[Search or retrieve standards]

Hold "Alt" / "Option" to enable pan & zoom

Use list_frameworks to discover candidates, then pin an exact snapshotId when reproducibility matters. Use get_capabilities before choosing code search, because a package's code coverage does not by itself authorize code_exact or code_prefix.

List accepted framework snapshots

list_frameworks returns accepted immutable snapshots rather than raw directories or unvalidated package candidates.

A minimal request is:

{
  "request": {}
}

The default page limit is 25; the maximum is 100.

A filtered discovery request can look like:

{
  "request": {
    "graphTypes": ["academic_standards"],
    "jurisdictions": ["Ghana"],
    "subjects": ["Mathematics"],
    "validationStatus": ["passed"],
    "limit": 25
  }
}

Available discovery filters

Field	Purpose
`query`	Free-text discovery across selected framework metadata
`graphTypes`	Require at least one requested graph type
`isCurrent`	Restrict to current or non-current snapshots
`issuingAuthorities`	Filter by source issuing authority
`jurisdictionTypes`	Filter by source jurisdiction type
`jurisdictions`	Filter by jurisdiction
`languages`	Filter by source language tags
`localGrades`	Filter by exact source-facing grades or stages
`normalizedGrades`	Filter by normalized retrieval facets
`subjects`	Filter by source-authored local subject
`validationStatus`	Filter by accepted package validation status
`limit`	Maximum snapshots returned on the page, from 1 through 100
`cursor`	Opaque continuation cursor from the preceding page

Filter values are normalized for matching without replacing the retained source values. Within one populated filter, any requested value may match. Across different populated filter dimensions, every dimension must match.

For example, requesting:

{
  "request": {
    "jurisdictions": ["Ghana", "Nigeria"],
    "subjects": ["Mathematics"]
  }
}

means Ghana OR Nigeria, and Mathematics.

Free-text discovery

The query field is for framework-catalog discovery, not standards search. Every whitespace-delimited query token must occur in the normalized concatenation of selected framework metadata such as framework ID, snapshot ID, authority, jurisdiction, subject, name, provider, or source version.

For standards content, use search_standards instead.

Read framework identity carefully

Each returned item represents one immutable snapshot and includes source metadata plus its accepted graph packages. Keep these identities distinct:

Identity	Meaning
`frameworkId`	Stable conceptual framework family
`snapshotId`	Immutable version of that framework
`graphPackageId`	Validated runtime delivery unit
`profileId` + `profileVersion`	Interpretation contract bound to the package

A framework may eventually have more than one accepted snapshot. Omitting snapshotId from later calls uses the catalog's unique-current routing rule; supplying it selects an exact immutable version.

Pin snapshots for reproducible work

For audits, evaluations, stored analyses, or published examples, retain both the frameworkId and snapshotId returned by discovery and pass the exact snapshot to downstream tools.

Retrieve one framework snapshot

Once the framework is known, use get_framework for the complete selected snapshot:

{
  "request": {
    "frameworkId": "ghana-nacca-primary-mathematics-basic-4-6"
  }
}

For an exact version:

{
  "request": {
    "frameworkId": "ghana-nacca-primary-mathematics-basic-4-6",
    "snapshotId": "ghana-nacca-primary-mathematics-basic-4-6@2019+43d21a2cb010"
  }
}

get_framework exposes exact source metadata, graph-package identity, counts, validation, rights, and declared package capabilities.

Inspect implemented capabilities

Call get_capabilities with no arguments before selecting a package-sensitive mode:

{}

The result reports the fixed public MCP surface and, for every accepted graph package, its authoritative implementedSearchModes.

Code coverage is not search authorization

A package may contain some coded nodes and still not implement every code-search mode. Use packages[].implementedSearchModes from get_capabilities before calling code_exact or code_prefix.

The current server intentionally reports unavailable features as well, including capabilities such as semantic search or persisted official alignment when they are not implemented.

Inspect structural statistics

get_framework_statistics is useful when you need to understand package shape before writing a traversal or evaluation request.

{
  "request": {
    "frameworkId": "india-cbse-science-learning-framework-classes-9-10",
    "graphType": "academic_standards"
  }
}

It reports deterministic counts including:

total item nodes and relationships;
coded and uncoded items;
local and normalized grade counts;
local and normalized statement-type counts;
hierarchy depth;
multi-parent target counts; and
unresolved relationship counts.

These are structural statistics. They do not establish difficulty, progression, coverage quality, or curriculum equivalence.

Pagination

list_frameworks uses checksum-protected opaque cursors. When hasMore is true, the tool returns continuation data containing a complete nextRequest.

Submit that nextRequest unchanged. Only the cursor differs from the preceding request. Changing the limit or any filter invalidates the cursor because pagination is bound to both the accepted catalog state and the exact cursor-free request.

Do not construct or edit cursors

Cursors are transport tokens, not user-visible offsets. Copy them exactly or use the complete nextRequest supplied by the tool.

Preserve local terminology

Framework discovery returns source-facing labels such as BASIC 1, PRIMARY THREE, P3, Class-5, and Class IX alongside normalized retrieval facets.

Use the source label when describing the curriculum. A normalized grade is useful for filtering but does not assert official equivalence between grade systems.

Natural-language example

An MCP host can perform the same workflow from a bounded instruction:

Use the curriculum-knowledge-graph connector to find the accepted Ghana Mathematics
framework. Preserve its exact framework ID, snapshot ID, source title, local grades,
adoption status, validation status, and graph-package identity. Then use
get_capabilities to report the search modes actually implemented for that package.
Do not infer grade equivalence or capabilities that are not explicitly reported.

When discovery is complete

You have enough information to proceed when you know the exact framework or snapshot, the source-facing grades relevant to the task, the package's implemented search modes, and any structural caveats that could affect retrieval or traversal.

Next: Search and retrieve standards