Skip to main content

Learning Overview

[!FROZEN] MPLP Protocol v1.0.0 Frozen Specification Freeze Date: 2025-12-03 Status: FROZEN (no breaking changes permitted) Governance: MPLP Protocol Governance Committee (MPGC) License: Apache-2.0 Note: Any normative change requires a new protocol version.

1. Scope & Position​

1.1 What are Learning Feedback Duties?​

Learning Feedback Duties define the protocol-level obligations for LearningSample data structures and collection points. They establish:

  • What format learning samples should use (schemas)
  • When to collect samples (recommended triggers)
  • What data each sample should contain (input/state/output/meta)

Critical Distinction: Learning Feedback Duties define DATA FORMAT obligations, NOT training behavior requirements.

1.2 Purpose​

Convert project execution history into structured learning samples that can be used for:

  • Model Training: Fine-tuning LLMs on domain-specific patterns
  • Evaluation: Assessing model performance on real-world scenarios
  • Counterfactual Reasoning: "What if" analysis on historical decisions
  • Quality Improvement: Identifying patterns in successful vs failed executions

2. Relationship with Other Layers​

2.1 Layer Hierarchy​

  • L1/L2 (Schemas + Modules): Define Context, Plan, Confirm, Trace, Role, Collab, etc.
  • Profiles (SA/MAP): Define single/multi-agent execution semantics
  • Observability: Define structured event streams during execution
  • Learning Feedback: Compress/abstract objects + events into LearningSamples for offline learning

2.2 Data Flow​

Execution [L2 Objects + Events] LearningSamples Training/Evaluation

LearningSamples reference L2 objects and Observability events via IDs, not embed

them.

3. LearningSample Concept​

3.1 What is a LearningSample?​

A LearningSample represents one "learnable unit of experience," such as:

  • How a user intent was clarified and resolved
  • The impact of a delta change and compensation
  • A pipeline stage's success/failure pattern
  • A confirm decision's approval/rejection reasoning
  • PSG evolution during a session
  • Multi-agent coordination performance

3.2 Core Structure​

Every LearningSample has 4 sections:

SectionPurposeRequired
inputAbstracted input conditions, intent, context summaryYes
stateSystem state snapshot before execution (PSG, config, etc.)Optional
outputActual execution results, decisions, changesYes
metaLabels, quality signals, provenance IDs, human feedbackOptional

Design Principle: Keep samples abstracted and compressed. Don't duplicate entire L2 objects; use summaries and references.

4. LearningSample Families​

MPLP v1.0 defines 6 LearningSample families:

4.1 Intent Resolution​

Purpose: Capture how user/business intent is clarified and converted to executable plans

Typical Sources:

  • Plan.intent_model
  • Dialog.turns
  • IntentEvent, DeltaIntentEvent

Example Use Case: Train model to better understand ambiguous user requests

4.2 Delta Impact​

Purpose: Capture delta change effects and impact analysis outcomes

Typical Sources:

  • Delta intents
  • Impact analysis artifacts
  • Compensation plans
  • ImpactAnalysisEvent, CompensationPlanEvent

Example Use Case: Learn to predict change ripple effects accurately

4.3 Pipeline Outcome​

Purpose: Capture pipeline stage-level success/failure patterns

Typical Sources:

  • PipelineStageEvent
  • Trace module
  • RuntimeExecutionEvent

Example Use Case: Predict which pipeline configurations succeed/fail

4.4 Confirm Decision​

Purpose: Capture approval/rejection decisions and reasoning

Typical Sources:

  • Confirm.decisions[]
  • RuntimeExecutionEvent

Example Use Case: Learn approval patterns for risky operations

4.5 Graph Evolution​

Purpose: Capture PSG structural changes over time

Typical Sources:

  • GraphUpdateEvent
  • PSG snapshots

Example Use Case: Understand healthy vs problematic graph evolution

4.6 Multi-Agent Coordination​

Purpose: Capture SA/MAP collaboration patterns and performance

Typical Sources:

  • MAP Events (MAPSessionStarted, etc.)
  • Collab + Network modules

Example Use Case: Optimize agent role distribution and coordination

5. Compliance Boundaries​

5.1 What is REQUIRED for v1.0 Compliance?​

Schema Stability: LearningSample schemas (core + family-specific) MUST be stable and backward-compatible.

Schema Conformance: IF a runtime emits LearningSamples, they MUST conform to MPLP schemas.

Sample Collection: Runtimes SHOULD emit LearningSamples at recommended collection points, but v1.0 compliance does NOT mandate this.

Training Behavior: Where samples are stored, how models are trained, and retraining frequency are ENTIRELY product implementation decisions.

5.3 Summary​

Aspectv1.0 Status
Schema definitions existREQUIRED
Schemas are stableREQUIRED
Runtime emits samplesRECOMMENDED
Samples conform to schemas (when emitted)REQUIRED
Training/storage strategyOut of scope

6. Collection Timing​

6.1 When to Generate LearningSamples?​

General Principle: Generate samples at decision boundaries and outcome observations.

Recommended Triggers:

  • After successful intent plan conversion
  • When impact analysis completes
  • On pipeline stage completion (success/failure)
  • When confirm decision finalized
  • On major PSG topology changes
  • At MAP session completion

See: Learning Collection Points for detailed specifications

7. Storage and Lifecycle (Out of Scope)​

MPLP v1.0 does NOT specify:

  • Where to store LearningSamples (filesystem, database, S3, etc.)
  • How long to retain samples
  • When to trigger training
  • What models to train

These are product implementation decisions for TracePilot, Coregentis, or other runtimes.

8. Privacy and Security Considerations​

Important: LearningSamples may contain sensitive information:

  • User requests
  • Business logic
  • Architectural decisions
  • Performance metrics

Recommendations (not protocol requirements):

  • Implement PII scrubbing before storage
  • Apply access controls to sample repositories
  • Consider differential privacy for sensitive domains
  • Document data retention policies

9. References​

Core Documentation:

Schemas:

  • schemas/v2/learning/mplp-learning-sample-core.schema.json
  • schemas/v2/learning/mplp-learning-sample-intent.schema.json
  • schemas/v2/learning/mplp-learning-sample-delta.schema.json

Examples:

  • examples/learning/flow-01-intent-sample.json
  • examples/learning/flow-05-confirm-sample.json

Compliance:

End of MPLP Learning Feedback Duties Overview

Learning Feedback Duties establish the protocol-level foundation for converting execution history into structured learning samples, enabling future model training and continuous improvement without mandating specific training strategies.​

2025 Bangshi Beijing Network Technology Limited Company Licensed under the Apache License, Version 2.0.