compress-memory

Extract lessons from training documents and update KNOWLEDGE_BASE.md.

Updated Mar 23, 2026

Search Tags:#workflow #compression #archive #training #knowledge-management #knowledge-base #director-approval

Ready to run

npx skills add https://github.com/swishstar/vwpm-assistant --skill compress-memory

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: compress-memory
Source: https://github.com/swishstar/vwpm-assistant/tree/main/.cursor/skills/compress-memory
Command: npx skills add https://github.com/swishstar/vwpm-assistant --skill compress-memory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill reduces the active context window by aggregating lessons learned and business preferences from training documents into KNOWLEDGE_BASE.md, then archives the raw sources to preserve provenance.

Core Features & Use Cases

Inventory training/ contents and identify extractable knowledge such as business preferences, corrections, recurring patterns, and named entities.
Merge new insights into KNOWLEDGE_BASE.md without duplicating existing entries, updating the last compressed timestamp, and presenting changes for director approval.
Archive processed training files to archive/ after green-light approval to maintain a lean working context.

Quick Start

Instruct the system to run maintenance to compress memory by extracting insights from training/ and updating KNOWLEDGE_BASE.md, then archive the raw files.

Frequently Asked Questions about compress-memory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I compress training documents into a knowledge base?▼

To compress training documents into a knowledge base, extract lessons learned and business preferences from the training directory, merge them into KNOWLEDGE_BASE.md with deduplication, and archive the raw inputs after extraction.

How does knowledge base deduplication work when archiving training files?▼

Knowledge base deduplication works by identifying extractable insights from training files and merging new entries into KNOWLEDGE_BASE.md without duplicating existing records, ensuring a lean context window before archiving raw sources.

Do I need director approval to update KNOWLEDGE_BASE.md?▼

Yes, you need director approval to update KNOWLEDGE_BASE.md. The workflow implements a green-light approval step to present extracted changes before writing to the knowledge base and archiving processed training files.

When should I run knowledge base maintenance for training documents?▼

You should run knowledge base maintenance for training documents during weekly intervals or upon director-initiated commands to actively reduce the context window by aggregating new lessons learned and business preferences.

What is the best way to archive raw training files after knowledge extraction?▼

The best way to archive raw training files after knowledge extraction is to move them to an archive directory after receiving green-light approval, preserving provenance while maintaining a lean working context in the knowledge base.