Claude-skill-registry-data meta-prompt-framework

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry-data
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry-data "$T" && mkdir -p ~/.claude/skills && cp -r "$T/data/meta-prompt-framework" ~/.claude/skills/majiayu000-claude-skill-registry-data-meta-prompt-framework && rm -rf "$T"
manifest: data/meta-prompt-framework/SKILL.md
source content

Meta-Prompt Framework Skill

This skill implements a systematic approach to prompt engineering and task execution optimization.

When This Skill Activates

  • Designing prompts for complex tasks
  • Implementing multi-phase workflows
  • Optimizing LLM task execution
  • Creating role-specific instructions
  • Building iterative refinement loops

Three-Phase Task Flow

Phase 1: Initialization

objectives:
  - Analyze task context
  - Clarify overall goal
  - Generate initial outline
  - Prepare inputs for processing

deliverables:
  - Task interpretation
  - Requirements list
  - Preliminary approach

instruction: |
  Prepare inputs for deeper processing and draft
  an initial outline of the solution structure.

Phase 2: Refinement

objectives:
  - Refine initial outputs
  - Identify inconsistencies
  - Explore alternatives
  - Ensure modularity

deliverables:
  - Quality-reviewed output
  - Alternative approaches
  - Edge cases identified

instruction: |
  Refine the output to meet quality standards,
  considering modularity and adaptability.

Phase 3: Completion

objectives:
  - Finalize against criteria
  - Review constraints
  - Verify structure
  - Generate final output

deliverables:
  - Complete output
  - Compliance check
  - Quality metrics

instruction: |
  Generate the final output, ensuring clarity,
  accuracy, and adherence to the task flow.

Success Criteria

CriterionCheck
AccuracyFactual correctness verified
ConsistencyReplicable across executions
ClaritySimple and understandable
EfficiencyCompleted in reasonable time

Refinement Strategies

Evaluate-Apply-Feedback Loop

  1. Evaluate outputs for accuracy/consistency
  2. Apply corrections and improvements
  3. Generate feedback for context update
  4. Iterate until criteria met

Context Evolution

  • Update context files with new insights
  • Document improvements for future reference
  • Feed learnings back into prompts

Role Separation

Task Executor

  • Generates outputs following task flow
  • Implements three-phase model
  • Focuses on deliverable creation

Task Reviewer

  • Reviews for clarity and accuracy
  • Checks meta-directive adherence
  • Provides improvement suggestions

Templates

Phase Template

## Phase [N]: [Name]

**Objective:** [What to accomplish]

**Inputs:**
- [Input 1]
- [Input 2]

**Process:**
1. [Step 1]
2. [Step 2]

**Outputs:**
- [Deliverable 1]
- [Deliverable 2]

**Success Criteria:**
- [ ] [Criterion 1]
- [ ] [Criterion 2]

Meta-Directive Template

meta_directives:
  [directive_name]:
    priority: [high|medium|low|critical]
    action: "[specific action to take]"
    constraint: "[limitation to observe]"

Best Practices

  1. Phase Gates: Complete each phase before proceeding
  2. Explicit Criteria: Define success metrics upfront
  3. Role Clarity: Separate execution from review
  4. Iteration Limits: Set maximum refinement cycles
  5. Context Preservation: Maintain state across phases