Skip to content

m3dcodie/LLM-Capability-Framework-LCF

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🏗️ LLM Capability Framework (LCF)

A 6-Layer Protocol for Strategic AI Workflows & Career Architecture.

The LLM Capability Framework (LCF) is an open-source architectural pattern designed to solve the "Contextual Drift" and "Buzzword Trap" of modern AI workflows. By decoupling complex tasks into distinct functional layers, the LCF transforms raw technical data into high-altitude, Principal-grade professional artifacts.


🏛️ The 6-Layer Architecture

LCF categorizes every AI task into a specific layer, matching requirement depth to optimal model classes.

Layer Role Purpose Ideal Model
L1 The Scout Deterministic data extraction (JSON). 4B-14B Distilled
L2 The Engineer Semantic transformation & the "Mirror Test." 8B-32B General
L3 The Strategist Gap analysis & Judgment Alignment (JA). 14B-70B Logic-heavy
L4 The Researcher ROI Benchmarking & RAG-based expansion. Search-Augmented Agents
L5 The Director Artifact assembly & ATS Optimization. Frontier (Claude/GPT-5)
L6 The Arbiter Recursive logic audit & hallucination guarding. o1/o3 Reasoning models

Why LCF?

Modern professionals often undersell their impact because they lack access to high-level strategic benchmarks. LCF "heals" professional narratives by:

  1. Quantifying Impact: Bridging technical acts with market-verified ROI.
  2. Eliminating Hallucinations: Using the L6 Arbiter to cross-reference synthesized narratives against raw source truth.
  3. Strategic Positioning: Shifting profiles from "Task-Oriented" to "Value-Oriented."

How to Use

1. Extract & Map (L1-L2)

Input your raw CV or LinkedIn profile. The L1 Scout extracts data into a clean JSON schema, while the L2 Engineer standardizes the narrative.

2. Audit & Expand (L3-L4)

Run the Layer 3 Strategic Audit to identify "Principal Gaps." Use the Layer 4 Researcher to inject real-world ROI metrics and industry benchmarks.

3. Compose & Verify (L5-L6)

Generate your final Executive CV at Layer 5. Finally, invoke the Layer 6 Arbiter to ensure every claim is 100% defensible and grounded in your original data.


📂 Repository Structure

LLM-Capability-Framework-LCF/
├── 1_Capability_Layers/
│   ├── Layer_1_extraction.md
│   ├── Layer_2_translator.md
│   ├── Layer_3_interpretation.md
│   ├── Layer_4_expansion.md
│   ├── Layer_5_composition.md
│   └── Layer_6_agency.md
├── 2_Architectural_Patterns/
│   ├── scout_auditor_pattern/
│   └── sub_question_retrieval/
├── 3_Evaluation_Benchmarks/
│   ├── Layer_1_extraction_accuracy_tests/
│   │   ├── prompt.md
│   │   ├── analysis.md
│   │   ├── claude-haiku-4-5-results.json
│   │   └── gemini-3-pro-results.json
│   ├── Layer_2_the_translator_tests/
│   │   ├── 1_prompt.md
│   │   ├── 2_mirror-test-prompt.md
│   │   ├── gemini-3-flash-output.md
│   │   ├── gemini3_flash_results.md
│   │   ├── haiku_mirror_audit.md
│   │   ├── qwen3-max--gemini3-flash-analysis.md
│   │   ├── qwen3-max-2025-09-23-output.md
│   │   └── qwen3-max-2025-09-23-results.md
│   ├── Layer_3_interpretation_tests/
│   │   ├── 1_main_prompt.md
│   │   ├── 2_Qwen3-32B_response_1_main_prompt.md
│   │   ├── 3_mistral-large-3_response_1_main_prompt.md
│   │   ├── 4_analysis_Qwen3-32B_mistral-large-3_response.md
│   │   ├── 5_peer_review_prompt_contract.md
│   │   ├── 6_deepseek_r1_response_peer_review_prompt_contract.md
│   │   └── 7_analysis_L3.md
│   ├── L4_expansion_tests/
│   │   ├── 1_main_prompt.md
│   │   ├── 2_main_prompt_GPT_52_response.md
│   │   ├── 3_alex_rivera_transformation.md
│   │   └── 4_analysis.md
│   ├── Layer_5_composition_tests/
│   │   ├── 1_prompt.md
│   │   ├── 1_prompt_GPT_52_response.md
│   │   └── 3_prompt_GeminiPro_response.md
│   └── Layer_6_agency_6.md/
│       ├── 1_prompt_positive.md
│       ├── 1_prompt_negative.md
│       ├── Layer_6_DeepSeekR1_positive_case_response.md
│       └── Layer_6_DeepSeekR1_negative_case_response.md
├── CHANGELOG.md
├── LICENSE
├── README.md
├── llms.txt
└── llms-full.txt

👤 Author

M Suhail Tahir

LinkedIn | GitHub

About

Precision-Matched Intelligence. The LLM Capability Framework (LCF) is a structural standard for mapping specific task requirements to the most effective model architecture, token strategy, and reasoning depth.

Topics

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors