forgetful-encode-repo

Encoding a repository into a knowledge base involves bootstrapping project entities, memories, and documents directly from the codebase. This process requires MCP client tools and Bash to automate the extraction and structuring of codebase knowledge for AI agents.

Repository encoding is the process of storing decisions, conventions, and patterns as atomic memories linked to their source documents. It maintains provenance tracking to record the originating source code for each piece of knowledge extracted from the codebase.

You can refresh an existing project knowledge base by re-encoding the repository. This updates the existing entities and atomic memories to ensure that all recent changes in the codebase are accurately reflected in the shared knowledge base.

Yes, you need MCP client tools to execute the repository encoding process. The Skill relies on specific MCP dependencies for discovering, learning, and executing tools, alongside Bash for scripting, to successfully build the knowledge base.

The best way to track codebase provenance is by encoding the repository into a knowledge base that maintains a record of the source code and files for each extracted piece of knowledge. This links atomic memories directly to their respective documents.