mcp_knowledge 0.2.0 copy "mcp_knowledge: ^0.2.0" to clipboard
mcp_knowledge: ^0.2.0 copied to clipboard

Unified knowledge management system integrating facts, skills, profiles, and operations with facades, bridges, and event-driven architecture for the MCP ecosystem.

0.2.0 - 2026-04-28 - Orchestrator-Only Redesign #

Changed (breaking) #

  • mcp_knowledge is now orchestrator-only — no port definitions or wrappers; composes runtimes from mcp_fact_graph / mcp_skill / mcp_profile / mcp_philosophy / mcp_knowledge_ops.
  • KnowledgeSystem constructor accepts five optional runtimes (factGraph, skillRuntime, profileRuntime, philosophyEngine, opsRuntime). Partial orchestration supported via null.
  • KnowledgePorts is a flat 32-field container of mcp_bundle standard ports (legacy CollectionStoragePort removed).
  • Facade methods (SkillFacade.execute, ProfileFacade.apply, FactFacade, PhilosophyFacade) delegate to real runtimes; 0.1.0 stand-ins are gone.
  • New dependency: mcp_bundle ^0.3.0.

Added #

  • OpsFacade wrapping OpsRuntime's workflow / pipeline / runbook ports (replaces BundleFacade).
  • KnowledgeSystem.stub() convenience factory.

Removed #

  • Six bridges (skill_fact, profile_fact, bundle_system, philosophy_skill/profile/fact) and five raw-accessor adapters — replaced by Contract Layer port wiring inside the runtimes.
  • BundleFacade — bundle deployment is now an OpsRuntime workflow concern.
  • KnowledgeSystem.runCuration / runSummarization / runPatternMining / loadBundle / executeSkill / applyProfile / executePipeline / executeWorkflow / resumeFromCheckpoint — delegated to system.ops.* / system.skill.* / system.profile.*.

0.1.0 Initial Release #

Added #

Core Features

  • KnowledgeSystem

    • Unified entry point for all knowledge operations
    • Configuration management with KnowledgeConfig
    • Port-based dependency injection with KnowledgePorts
    • Graceful shutdown and lifecycle management
  • Facades

    • FactFacade for simplified fact operations
    • SkillFacade for skill execution
    • ProfileFacade for profile rendering
    • BundleFacade for bundle management
  • Bridges

    • SkillFactBridge for connecting skill outputs to fact graph
    • ProfileFactBridge for injecting fact context into profiles
    • BundleSystemBridge for deploying bundles across subsystems
  • Event System

    • KnowledgeEventBus for event distribution
    • Typed event streams with on<T>() method
    • Event subscription with handlers
  • Events

    • FactConfirmedEvent - when a candidate becomes a fact
    • CandidateCreatedEvent - when a new candidate is created
    • SummaryRefreshedEvent - when a summary is updated
    • SkillExecutedEvent - when a skill completes
    • ClaimsRecordedEvent - when claims are recorded
    • ProfileRenderedEvent - when a profile is rendered
    • BundleLoadedEvent - when a bundle is deployed
    • PipelineCompletedEvent - when a pipeline finishes
    • SystemShutdownEvent - when system shuts down
  • Configuration

    • KnowledgeConfig with hierarchical settings
    • FactGraphConfig for fact graph settings
    • SkillConfig for skill execution settings
    • ProfileConfig for profile rendering settings
    • SchedulerConfig for scheduling settings
    • EventConfig for event system settings
    • LoggingConfig for logging settings
    • FeatureFlags for feature toggles
    • Preset configurations: defaults, development, production
  • Ports

    • KnowledgePorts unified port container
    • StoragePort for data persistence
    • LlmPort for LLM integration
    • McpPort for MCP communication
    • EvidencePort for evidence processing
    • ExpressionPort for template evaluation
    • Stub implementations for testing
  • Adapters

    • SkillPortsAdapter for skill port conversion
    • ProfilePortsAdapter for profile port conversion
    • FactGraphPortsAdapter for fact graph port conversion

Integration #

  • Integrates mcp_bundle for bundle management
  • Integrates mcp_fact_graph for knowledge storage
  • Integrates mcp_skill for AI capabilities
  • Integrates mcp_profile for AI personas
  • Integrates mcp_knowledge_ops for operations

Architecture #

  • Facade pattern for simplified APIs
  • Bridge pattern for cross-package integration
  • Event-driven architecture for loose coupling
  • Port-based dependency injection

Support and Contributing #

0
likes
0
points
211
downloads

Publisher

unverified uploader

Weekly Downloads

Unified knowledge management system integrating facts, skills, profiles, and operations with facades, bridges, and event-driven architecture for the MCP ecosystem.

Homepage
Repository (GitHub)
View/report issues

Topics

#knowledge #mcp #ai #dart #integration

License

unknown (license)

Dependencies

mcp_bundle, mcp_fact_graph, mcp_knowledge_ops, mcp_philosophy, mcp_profile, mcp_skill, meta

More

Packages that depend on mcp_knowledge