CAKE (Context-Aware Knowledge Enhancement)

A GitHub template repository designed to enhance AI assistant interactions with context-aware prompts and organized knowledge management for development teams.

Overview

This template addresses common limitations of pre-trained language models in enterprise environments, including outdated training data and lack of domain-specific knowledge such as company acronyms, team terminology, and project-specific context.

Without CAKE - Generic Responses

%%{init: {'theme':'base', 'themeVariables': { 'fontSize': '14px', 'fontFamily': 'arial', 'background': '#ffffff' }}}%%
graph TD
    A1["Question:<br/>'How should I structure this API endpoint?'"] --> B1["Standard REST Principles"]
    B1 --> C1["'Use HTTP verbs correctly<br/>Return proper status codes<br/>Use JSON responses'"]
    C1 --> D1["Doesn't Match:<br/>Company API gateway<br/>Authentication patterns<br/>Response formats"]
    
    style A1 fill:#ffffff,stroke:#333,font-size:12px
    style B1 fill:#ffffff,stroke:#333,font-size:12px
    style C1 fill:#ffffff,stroke:#333,font-size:12px
    style D1 fill:#ff9999,stroke:#333,font-size:12px

Loading

With CAKE - Context-Aware Responses

%%{init: {'theme':'base', 'themeVariables': { 'fontSize': '14px', 'fontFamily': 'arial', 'background': '#ffffff' }}}%%
graph TD
    A2["Question:<br/>'How should I structure this API endpoint?'"] --> B2["CAKE System"]
    B2 --> P["Check prompt.md"] --> K["Check /knowledge folder"] --> B2
    P --> PERSONAL["Check /personal folder"] --> B2
    B2 --> C2["API Gateway Config"]
    B2 --> C3["Auth Service Patterns"]
    B2 --> C4["Response Templates"]
    C2 --> D2["Contextual Guidance"]
    C3 --> D2
    C4 --> D2
    D2 --> E2["'Use /api/v2/ prefix<br/>Implement JWT middleware<br/>Follow ResponseWrapper<T> pattern<br/>Include correlationId header'"]
    E2 --> F2["Consistent with<br/>Existing Services"]
    
    style A2 fill:#ffffff,stroke:#333,font-size:12px
    style B2 fill:#ffffff,stroke:#333,font-size:12px
    style C2 fill:#e6f3ff,stroke:#333,font-size:12px
    style C3 fill:#e6f3ff,stroke:#333,font-size:12px
    style C4 fill:#e6f3ff,stroke:#333,font-size:12px
    style D2 fill:#ffffff,stroke:#333,font-size:12px
    style E2 fill:#ffffff,stroke:#333,font-size:12px
    style F2 fill:#99ff99,stroke:#333,font-size:12px

Loading

Inspired by Burke Holland's prompt engineering work: https://gist.github.com/burkeholland

Repository Structure

0. Project Repositories (repos/)

Add your Git repositories here to provide code context to AI assistants. For multi-repository services (e.g., service-api and service-frontend), organize them under a common folder structure. This enables AI agents to interactively navigate and understand the broader project context.

Example structure:

repos/
├── service_1/
├── service_2-mobile/
├── service_3-api/
├── service_4-backend/
├── raw/
│   ├── architecture/
│   │   ├── design-docs/
│   │   └── system-diagrams/
│   ├── meetings/
│   │   └── team-notes/
│   └── specifications/
│       ├── technical-specs/
│       └── requirements/
├── knowledge/
│   ├── company/
│   │   ├── acronyms.md
│   │   └── terminology.md
│   ├── processes/
│   │   └── workflows.md
│   └── standards/
│       └── guidelines.md
└── personal/
    ├── daily/
    │   └── YYYY-MM-DD.md
    ├── learning/
    │   └── til.md
    └── tasks/
        └── in-progress.md

1. Raw Knowledge (raw/)

Store unprocessed documentation in Markdown (.md) or text (.txt) format. This serves as a knowledge dump for:

• Project design documents
• Team meeting notes
• Technical specifications
• Architecture decisions
• Requirements documents

Purpose: Central repository for all project-related documentation that needs to be processed into structured knowledge.

2. Processed Knowledge (knowledge/)

AI-curated and organized information extracted from the raw/ folder. This folder contains:

• Company-specific terminology and acronyms
• Team processes and workflows
• Project context and domain knowledge
• Technical standards and guidelines

Workflow: Use your preferred LLM (e.g., Claude Sonnet 4.0) with the provided prompt.md instructions to automatically process raw documents into structured knowledge. The AI will read the raw/ folder contents and create organized, searchable knowledge files.

3. Personal Workspace (personal/) [Optional]

Track individual work progress and maintain context continuity:

• Daily summaries of completed tasks
• Work-in-progress tracking
• Technical learnings (TIL - Today I Learned)
• Task continuation notes for next-day pickup

Benefits:

• Reduces context switching overhead
• Provides performance boost for similar/repeated work
• Maintains historical context for decision-making
• Enables better task prioritization and planning

Usage Workflow

1. Initialize: Clone this template and populate the repos/ folder with your project repositories
2. Collect: Add all relevant documentation to the raw/ folder
3. Process: Use AI to transform raw documents into structured knowledge using the provided prompts
4. Maintain: Regularly update both raw and processed knowledge as projects evolve
5. Personalize: Optionally use the personal/ folder for individual productivity tracking

Benefits

• Enhanced Context: AI assistants gain deep understanding of your specific projects and terminology
• Reduced Onboarding: New team members can quickly understand project context
• Improved AI Responses: More accurate and relevant assistance tailored to your environment
• Knowledge Preservation: Centralized documentation that evolves with your projects
• Productivity Gains: Faster development cycles through better context awareness

Getting Started

1. Use this template to create a new repository
2. Add your project repositories to the repos/ folder
3. Populate the raw/ folder with existing documentation
4. Configure your AI assistant to use the processed knowledge for enhanced interactions

Contributing

This template is designed to be adapted to your team's specific needs. Feel free to modify the folder structure and processes to match your workflow and organizational requirements.