AI/Tasks/HeuristicsUpgrades/Heuristics_Phase02_PatternMatchingLayer_Implementation.txt
|
PHASE 02 IMPLEMENTATION PROMPT - PATTERN MATCHING LAYER
Goal: Implement a deterministic pattern-matching layer that emits bounded evidence contributions only. Inputs: - AI/Tasks/HeuristicsUpgrades/Heuristics_Phase02_PatternMatchingLayer_Strict.txt - Outputs from Phase 01 Required code work: - Implement typed pattern contracts including: pattern id, trigger condition, match strength, ambiguity flag, bounded contribution, override flag, conflict metadata. - Implement trigger evaluation for high-confidence, fallback, ambiguity, and explicit override-intent patterns. - Add deterministic conflict resolution when multiple patterns match. - Enforce maximum pattern count and maximum aggregate pattern contribution. - Ensure pattern output feeds score composition but never directly sets authorization/refusal/confirmation/sandbox outcome. Implementation constraints: - Keep pattern evaluation deterministic and auditable. - Do not modify cost, session reinforcement, or clarification template policy beyond interface compatibility. Minimum tests to add/update: - Trigger determinism tests. - Overlap/conflict resolution tests. - Ambiguity routing tests. - Tests proving pattern evidence cannot weaken safety gates. Validation commands: - dotnet build src/TechToolbox.Agent/TechToolbox.Agent.csproj -c Release - dotnet test src/TechToolbox.Agent/Tests/TechToolbox.Agent.Tests.csproj -c Release --filter "Pattern|Trigger|Ambiguity|Override|Fallback" Deliverable: Return pattern contracts, conflict rules, and test evidence for deterministic behavior. |