--- frontmattername: Core Patterns
description: Base templates for error escalation, decision criteria, and enforcement. Referenced by agents and commands to avoid duplication.
version: 2.0.0
Boundary tests required:
boundary_success_80: Exactly 80, yields success
boundary_warning_79: 78.5-79.9, yields warning
boundary_error_59: 58.5-59.9, yields error
</example>
Examples:
EXEC-B001: Execute command mandatory behavior 1
DEF-P002: Define command prohibited behavior 2
</example>
Use attribute values:
patterns: Shared templates (core-patterns, parallelization-patterns, workflow-patterns)
tools: Tool-specific usage patterns (serena-usage, context7-usage)
workflow: How-to guides and methodologies (investigation-patterns, execution-workflow)
domain: Domain knowledge and best practices (nix-ecosystem, typescript-ecosystem)
</example>
</pattern>
</patterns>
<anti_patterns>
<avoid name="inline_error_escalation">
<description>Duplicating full error_escalation in each file</description>
<instead>Reference core-patterns and customize only examples</instead>
</avoid>
<avoid name="inconsistent_thresholds">
<description>Using different confidence thresholds (75, 80, 85) across files</description>
<instead>Always use 60/80 boundaries as defined in core-patterns</instead>
</avoid>
<avoid name="missing_boundary_tests">
<description>Omitting boundary validation tests (59/60, 79/80)</description>
<instead>Always include boundary_success_80, boundary_warning_79, boundary_error_59 tests</instead>
</avoid>
<avoid name="weight_sum_mismatch">
<description>Decision criteria weights not summing to 1.0</description>
<instead>Verify weights sum to exactly 1.0 (e.g., 0.4+0.3+0.3)</instead>
</avoid>
<avoid name="inconsistent_behavior_ids">
<description>Using different ID formats across files</description>
<instead>Use PREFIX-TYPE-NUMBER format consistently</instead>
</avoid>
</anti_patterns>
<best_practices>
<practice priority="critical">Reference core-patterns for error_escalation, decision_criteria, enforcement templates</practice>
<practice priority="critical">Always include all 5 validation tests: success, boundary_success_80, boundary_warning_79, boundary_error_59, error</practice>
<practice priority="critical">Ensure decision criteria weights sum to 1.0</practice>
<practice priority="high">Customize error_escalation examples to be domain-specific while keeping structure</practice>
<practice priority="high">Use consistent behavior ID prefixes within each agent/command</practice>
</best_practices>
<rules priority="critical">
<rule>Always include all 5 validation tests for decision_criteria</rule>
<rule>Boundary tests must use exact threshold values (80, 79.x, 59.x)</rule>
<rule>Error escalation must have exactly 4 severity levels</rule>
<rule>Weights in decision_criteria must sum to 1.0</rule>
</rules>
<rules priority="standard">
<rule>Use refs tag to reference this skill from agents and commands</rule>
<rule>Customize examples in error_escalation while keeping structure</rule>
<rule>Use consistent behavior ID naming convention</rule>
</rules>
<enforcement>
<mandatory_behaviors>
<behavior id="CORE-B001" priority="critical">
<trigger>When creating new agent or command</trigger>
<action>Reference core-patterns skill in refs section</action>
<verification>refs tag contains core-patterns</verification>
</behavior>
<behavior id="CORE-B002" priority="critical">
<trigger>When defining decision_criteria</trigger>
<action>Include all 5 validation tests with boundary cases</action>
<verification>Tests include boundary_success_80, boundary_warning_79, boundary_error_59</verification>
</behavior>
</mandatory_behaviors>
<prohibited_behaviors>
<behavior id="CORE-P001" priority="critical">
<trigger>Always</trigger>
<action>Using non-standard confidence thresholds</action>
<response>Use 60/80 thresholds as defined in this skill</response>
</behavior>
</prohibited_behaviors>
</enforcement>
<error_escalation>
<level severity="low">
<example>Minor inconsistency in behavior ID format</example>
<action>Note in report, proceed</action>
</level>
<level severity="medium">
<example>Missing one boundary test</example>
<action>Document issue, add missing test</action>
</level>
<level severity="high">
<example>Decision criteria weights do not sum to 1.0</example>
<action>STOP, fix weight distribution before proceeding</action>
</level>
<level severity="critical">
<example>Error escalation missing critical level</example>
<action>BLOCK operation, require complete 4-level structure</action>
</level>
</error_escalation>
<related_skills>
<skill name="parallelization-patterns">Parallel execution strategies and timeout configuration</skill>
<skill name="workflow-patterns">Output formats, reflection checkpoints, agent weights</skill>
<skill name="serena-usage">For memory operations to store pattern decisions</skill>
</related_skills>
<constraints>
<must>Define exactly 4 severity levels for error_escalation</must>
<must>Include all 5 validation tests for decision_criteria</must>
<must>Ensure weights sum to 1.0</must>
<must>Use standard confidence thresholds (60/80)</must>
<avoid>Inventing new severity levels</avoid>
<avoid>Omitting boundary tests</avoid>
<avoid>Using non-standard thresholds</avoid>
</constraints>