MCP Profile

Positioning: mcp_profile is an internal component of the MakeMind knowledge stack exposed through the mcp_knowledge facade. Application code should import package:mcp_knowledge/mcp_knowledge.dart — the symbols declared here are re-exported from there. Direct package:mcp_profile/ imports remain valid for advanced or integration scenarios but are discouraged in product code.

AI persona definitions with template rendering, context injection, profile selection, and appraisal scoring. Pluggable engines (appraisal / decision / expression) drive a unified runtime that exposes the mcp_bundle standard ports.

Components

• Definition — Profile, sections, capability metadata, builder, registry, renderer.
• Expression — formatter, policy, style, evaluator (templated profile prompts).
• Appraisal — engine, metric definition / source, normalization, results.
• Decision — evaluator, decision policy, policy condition, guidance.
• Bundle — profile bundle spec.
• Engines (re-exported for custom implementations and tests) — AppraisalEnginePort, DecisionEnginePort, ExpressionEnginePort, EnginePorts container.
• Standard port adapters — MetricsPortAdapter, AppraisalPortAdapter, DecisionPortAdapter, ExpressionPortAdapter, ProfileSummariesPortAdapter implementing mcp_bundle Contract Layer.
• Runtime — unified ProfileRuntime orchestrating Appraisal → Decision → Expression.
• Feature modules — cache, versioning, concurrency, fact-graph integration.

Quick Start

import 'package:mcp_profile/mcp_profile.dart';

final runtime = ProfileRuntime(
  registry: ProfileRegistry(),
  engines: EnginePorts(
    appraisal: StubAppraisalEnginePort(),
    decision: DefaultDecisionEnginePort(),
    expression: PassthroughExpressionEnginePort(),
    facts: factsPort,
    summaries: summariesPort,
    llm: llmPort,
  ),
);

final result = await runtime.apply(profileId, RuntimeProfileContext(...));

Support

• Issue Tracker
• Discussions

License

MIT — see LICENSE.