Claude Code Plugins

Community-maintained marketplace

Feedback

Universal 7-phase execution pattern for all substantive work. USE WHEN substantive task, workflow execution, Core Algorithm, OBSERVE THINK PLAN BUILD EXECUTE VERIFY LEARN, OR execution pattern needed. The meta-skill that all other skills implement.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name CoreAlgorithm
description Universal 7-phase execution pattern for all substantive work. USE WHEN substantive task, workflow execution, Core Algorithm, OBSERVE THINK PLAN BUILD EXECUTE VERIFY LEARN, OR execution pattern needed. The meta-skill that all other skills implement.

CoreAlgorithm

THE Universal Execution Pattern for Kai

All work follows this algorithm. No exceptions. Skills, workflows, and custom workflows all implement this pattern.


Voice Notification

When executing the algorithm workflow, do BOTH:

  1. Send voice notification:

    curl -s -X POST http://localhost:8888/notify \
      -H "Content-Type: application/json" \
      -d '{"message": "Executing task using the Core Algorithm at EFFORT_LEVEL effort"}' \
      > /dev/null 2>&1 &
    
  2. Output text notification:

    Executing this task using the **Core Algorithm** at **[Effort Level]** effort...
    

Full documentation: ~/.claude/Skills/CORE/SkillNotifications.md

Workflow Routing

Workflow Trigger File
ExecuteCoreAlgorithm Substantive task detected by hook Workflows/ExecuteCoreAlgorithm.md

The Algorithm

OBSERVE → THINK → PLAN → BUILD → EXECUTE → VERIFY → LEARN

Quick Reference

The 7 Phases:

# Phase Purpose Output
1 OBSERVE Gather context Current state + constraints
2 THINK Generate hypotheses (min 3) Selected approach + verification plan
3 PLAN Design solution Execution specification
4 BUILD Create Test Plan Executable test criteria table
5 EXECUTE Implement Deliverables
6 VERIFY Execute Test Plan Pass/fail per criterion
7 LEARN Improve algorithm itself Algorithm adjustments

Effort Levels:

Level Time Agents Trigger Words
Fast <30s Kai only "quick", "just", "simple"
Standard <2min Up to 3 Default for most work
Deep <10min Up to 5 "think about", "explore"
Excellence Continuous Up to 16 "best possible", "critical"

Context Files

File Purpose
CoreAlgorithm.md Full algorithm definition, LEARN phase details, effort matrix
WorkflowExecution.md Detailed 7-phase execution guide, test plan structure
~/.claude/CustomWorkflows/README.md Custom workflow directory system

Examples

Example 1: Deep strategic plan

User: "Create a marketing plan for PAI"
→ Hook detects substantive task (Deep effort)
→ Invokes ExecuteCoreAlgorithm workflow
→ OBSERVE: Gather PAI context, platforms, audience
→ THINK: Generate 5+ marketing strategy hypotheses with agents
→ PLAN: Design campaign structure with Architect
→ BUILD: Define success criteria table (8+ criteria)
→ EXECUTE: Create the marketing plan with parallel Engineers
→ VERIFY: Check against criteria with QA agent
→ LEARN: Capture algorithm improvements

Example 2: Standard bug fix

User: "Fix the auth bug in login.ts"
→ Hook detects substantive task (Standard effort)
→ OBSERVE: Read relevant code
→ THINK: 3 hypotheses for root cause
→ PLAN: Fix approach
→ BUILD: Test plan with 5 criteria
→ EXECUTE: Implement fix
→ VERIFY: Run tests
→ LEARN: Note if algorithm pattern needs adjustment

Example 3: Fast simple edit

User: "Fix this typo"
→ Hook detects simple task (Fast effort)
→ Quick 7-phase cycle (mostly mental)
→ Execute the fix
→ Verify it works

Last Updated: 2025-12-31