Chunk text and documents into semantic pieces with metadata for Vector.Store.
npx skills add https://github.com/ShipFail/promptware --skill memory-manager-shipfail
Please help me install this Agent Skill. Skill: Memory Manager Source: https://github.com/ShipFail/promptware/tree/main/os/skills/memory-manager Command: npx skills add https://github.com/ShipFail/promptware --skill memory-manager-shipfail
Memory Manager automates data preparation for the PromptWar̊e ØS Vector Memory Subsystem, bridging the gap between raw files and vector storage by structuring data for embedding pipelines and retrieval.
To chunk a file, run deno run -A os/skills/memory-manager/chunker.ts --file README.md. To chunk arbitrary text, run deno run -A os/skills/memory-manager/chunker.ts "Some long text to chunk..." The default chunk size is 1000 characters; adjust with --size.
High-intent search queries and answers about installing and using this skill.
To prepare raw text documents for vector memory ingestion, you can use an automated chunker that splits files into semantic chunks under a configurable size, attaches source metadata, and outputs JSON or readable text for embedding workflows.
Semantic chunking for embedding pipelines is the process of partitioning input text into meaningful segments under a specific character limit. This ensures the data maintains context and structure while attaching metadata like source, index, and total for traceability.
You can chunk a file using Deno by running the chunker script with the file path argument. The script partitions the document into semantic chunks with a default size of 1000 characters, which can be adjusted using the size flag.
Yes, you can attach metadata to text chunks before sending them to a vector store. The chunking process automatically extracts and formats source, index, and total metadata to ensure traceability within the Vector Memory Subsystem.
Yes, the data preparation tool outputs JSON or readable text specifically formatted for vector stores. This ensures the chunked documents with attached metadata are ready for immediate ingestion into the Vector.Store system.
The default chunk size for text processing is 1000 characters. You can adjust this default size by configuring the size parameter when running the chunker script on your input files or arbitrary text.
gensparxGenerate visual and file-based diffs from text or patch inputs.
patrykkopycinskiSet up a RAG pipeline with Elasticsearch for vector retrieval.
Himanshu040604Developes RAG systems integrating vector databases and advanced retrieval strategies for LLM applications.
vaayneExtract text, tables, metadata, and images from 91+ document formats.
