fabrik-codek

What This Skill Does

Fabrik-Codek is a sophisticated personal cognitive architecture that transforms how you interact with your own development history and knowledge. Unlike standard RAG systems that treat all data as equal, Fabrik-Codek treats your digital footprint—code, transcripts, and session history—as a living, breathing knowledge base. It employs an 11-step extraction pipeline to build a hybrid retrieval system combining vector embeddings, knowledge graphs, and full-text search. The engine analyzes your personal competence levels across various technology stacks and domains, allowing it to adapt its tone and depth specifically to your current proficiency. It monitors interaction success without requiring manual feedback, enabling continuous self-correction and optimization of retrieval strategies.

Installation

To integrate Fabrik-Codek into your OpenClaw environment, ensure you have Ollama installed as the local inference engine. Run the installation command via the CLI: clawhub install openclaw/skills/skills/ikchain/fabrik-codek. Once installed, configure it within your openclaw.json or ~/.claude/settings.json file as an MCP server. For standard use, define the command and args as ["mcp"]. For remote or SSE-based setups, specify --transport sse and your preferred port. After configuration, initialize your knowledge base by running fabrik init, fabrik graph build, fabrik rag index, fabrik profile build, and fabrik competence build to establish your personal profile and competence map.

Use Cases

• Personalized Debugging: Receive solutions that account for your specific coding patterns, preferred libraries, and past architectural decisions.
• Technical Onboarding: Use the competence scoring to get explanations tailored to your familiarity with a new framework.
• Knowledge Retrieval: Retrieve complex relationship data between disparate projects through the integrated knowledge graph.
• Adaptive Documentation: Generate documentation that mirrors your personal style and technical standards.

Example Prompts

1. "Based on my previous projects, what is the best way to structure this new service's database schema?"
2. "Explain the current authentication implementation in this repository, focusing on why I chose this specific library."
3. "Summarize the decisions we made regarding state management in the last three sessions."

Tips & Limitations

• Privacy: Fabrik-Codek runs entirely locally; it makes zero outbound network calls, keeping your intellectual property safe.
• Resource Usage: Because it indexes your entire history, initial build times for the knowledge graph and vector store can be intensive; plan to run these builds during off-hours.
• Ollama Dependency: Ensure your local Ollama instance is running and has sufficient memory to handle both the embedding models and the LLM selected for reasoning.