mcp_fact_graph library

MCP Fact Graph - Knowledge representation with 4-layer architecture.

Layers:

  • L0: Evidence Layer (ingestion, extraction)
  • L1: FactGraph Layer (candidates, entities, events, views)
  • L2: ContextOps Layer (context bundles, summaries, claims)
  • L3: SkillOps Layer (patterns, skills, rubrics, evaluation)

Classes

AggregateResult
Aggregation result.
BoundingBox
Bounding box for image regions.
CachedGraphStorage
Cached graph storage wrapper.
Candidate
Candidate represents an assembled fact/event pending confirmation.
CandidateField
A field in a candidate with value and confidence.
CandidateQuery
Query for candidates.
CandidateStoragePort
Port for candidate storage operations.
Claim
Claim represents an assertion to be verified.
ClaimVerificationInput
Input for claim verification.
ClaimVerificationPort
Port for claim verification using LLM.
ClaimVerificationResult
Claim verification result.
ComputationMeta
Computation metadata.
ConfirmationResult
Result of confirming a candidate.
ContextBundle
ContextBundle represents a curated context for LLM operations.
ContextSegment
A segment of context content.
ContextService
Service for L2 ContextOps Layer operations.
ContextStoragePort
Port for context storage operations.
DimensionScore
Score for a single dimension.
Entity
Entity represents a resolved real-world object in the fact graph.
EntityMatch
Entity match result.
EntityResolutionInput
Input for entity resolution.
EntityResolutionPort
Port for entity resolution using LLM.
EntityResolutionResult
Entity resolution result.
EntityStoragePort
Port for entity storage operations.
EvaluationIssue
Issue found during evaluation.
EvaluationRun
EvaluationRun represents a completed evaluation.
Event
Event represents a confirmed fact or event in the fact graph.
EventEdge
Edge connection from an event.
EventQuery
Query for events.
EventStoragePort
Port for event storage operations.
Evidence
Evidence represents the original, immutable input data.
EvidenceInput
Input for evidence ingestion.
EvidencePort
Port for evidence ingestion from external sources.
EvidenceQuery
Query for evidence.
EvidenceReference
Reference to evidence used in verification.
EvidenceService
Service for L0 Evidence Layer operations.
EvidenceStoragePort
Port for evidence storage operations.
ExtractionConfig
Configuration for fragment extraction.
FactEdge
An edge connecting two nodes in the fact graph.
FactGraph
A graph structure for storing and querying facts.
FactGraphService
Service for L1 FactGraph Layer operations.
FactNode
A node in the fact graph representing a piece of knowledge.
FactSource
Source information for a fact.
FieldCondition
Field condition for pattern matching.
Fragment
Fragment represents an extracted piece of data from evidence.
FragmentExtractorPort
Port for fragment extraction.
GraphQuery
A fluent query builder for fact graphs.
GraphStorage
Abstract interface for graph storage.
LlmCapabilities
LLM capabilities.
LlmMessage
LLM message in conversation.
LlmPort
Port for LLM operations.
LlmRequest
LLM request.
LlmResponse
LLM response.
MemoryGraphStorage
In-memory graph storage.
PathResult
Path result for graph traversal.
Pattern
Pattern represents a discovered recurring pattern.
PatternCriteria
Pattern matching criteria.
PatternFrequency
Pattern frequency statistics.
PatternMatch
Pattern match result.
QueryFilter
Query filter types.
QueryResult<T>
Result of a graph query with pagination info.
QuerySort
Sort specification.
Rubric
Rubric represents evaluation criteria for skills.
RubricDimension
Evaluation dimension within a rubric.
ScoreLevel
Score level definition.
Skill
Skill represents an executable capability.
SkillExecutionConfig
Skill execution configuration.
SkillExecutionResult
Skill execution result.
SkillOpsService
Service for L3 SkillOps Layer operations.
SkillOpsStoragePort
Port for skill operations storage.
SkillStats
Skill execution statistics.
SkillTrigger
Skill trigger configuration.
SourceCapabilities
Capabilities of an evidence source.
SourceMetadata
Source metadata for evidence.
SourcePosition
Position in source content.
SummaryNode
SummaryNode represents a hierarchical summary unit.
SummaryPeriod
Summary time period.
TokenUsage
Token usage statistics.
UnresolvedIssue
An unresolved issue preventing candidate confirmation.
VerificationResult
Verification result.
VersionedGraphStorage
Graph storage with versioning support.
View
View represents an aggregated computation over events/facts.
ViewPeriod
View period.
ViewQuery
Query for views.
ViewStoragePort
Port for view storage operations.

Enums

CandidateStatus
Candidate status.
ClaimType
Claim types.
ConditionOperator
Condition operators.
ContextPurpose
Context purpose types.
EdgeType
Types of edges in the fact graph.
EntityStatus
Entity status.
EvalIssueType
Evaluation issue types.
EvaluationStatus
Evaluation status.
EvaluationType
Evaluation types.
EventEdgeType
Edge types for event connections.
EventStatus
Event status.
EvidenceRelation
Evidence relation to claim.
EvidenceSourceType
Types of evidence sources.
EvidenceStatus
Evidence processing status.
ExtractionMethod
Extraction methods.
FragmentStatus
Fragment status.
FragmentValueType
Types of fragment values.
IssueSeverity
Issue severity levels.
IssueType
Types of unresolved issues.
NodeType
Types of nodes in the fact graph.
PatternStatus
Pattern status.
PatternType
Pattern types.
PeriodGranularity
Period granularity levels.
PeriodType
Period types.
RubricStatus
Rubric status.
SegmentType
Context segment types.
SelectionStrategy
Context selection strategies.
SkillCategory
Skill categories.
SkillStatus
Skill status.
SourceType
Types of fact sources.
SummaryStatus
Summary status.
SummaryType
Summary types.
TrendDirection
Trend directions.
TriggerType
Trigger types.
VerificationStatus
Verification status.
VerificationVerdict
Verification verdict.
ViewStatus
View status.
WeightingStrategy
Weighting strategies for dimension scores.

Extensions

QueryableGraph on FactGraph
Extension to add query method to FactGraph.