memory-distill

Capture user feedback and distill it into durable memory files.

1|Updated Mar 31, 2026

Search Tags:#feedback #learning #memory #content-optimization #autocrew #writing-preferences

Ready to run

npx skills add https://github.com/alextangson/AutoCrew --skill memory-distill-alextangson

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: memory-distill
Source: https://github.com/alextangson/AutoCrew/tree/main/skills/memory-distill
Command: npx skills add https://github.com/alextangson/AutoCrew --skill memory-distill-alextangson

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This memory-distill skill captures user feedback (approvals, rejections, edits, or performance signals) and distills it into a durable memory store to guide future content creation.

Core Features & Use Cases

L1 Working memory: ~/.autocrew/MEMORY.md (≤120 lines) for brand profile, audience, and writing preferences.
L2 Detailed logs: ~/.autocrew/memory/ for archived digests and on-demand references.
Trigger-based learning: captures feedback whenever users approve, reject, edit, or share performance data across topics or posts.
Preference synthesis: converts feedback into explicit, reusable writing rules and templates.
Periodic distillation: consolidates learnings over time to adapt to evolving preferences.

Quick Start

Enable memory distillation by activating the autocrew workflow to capture feedback and consolidate learnings into durable memory.

Frequently Asked Questions about memory-distill

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

FAQPage Schema

How do I capture user feedback into durable memory for content optimization?▼

Memory distillation captures feedback into durable memory by applying a two-layer architecture and structured logging workflow to persist learnings, storing brand profiles and writing preferences in working memory while archiving detailed digests locally.

What is the best way to save writing preferences from user edits and rejections?▼

Saving writing preferences from edits and rejections is best handled by synthesizing feedback into explicit, reusable writing rules, periodically consolidating these learnings to adapt to evolving audience preferences over time.

How does two-layer memory architecture work for writing preference adaptation?▼

Two-layer memory architecture works by maintaining an L1 working memory file for active brand profiles and audience preferences, while an L2 directory archives detailed digests and logs for on-demand reference and periodic distillation.

Does memory distillation require specific dependencies to persist learning logs?▼

Memory distillation requires no external dependencies to persist learning logs, utilizing a structured workflow to write and consolidate feedback directly into local memory files within the designated configuration directory.

Can I use periodic distillation to adapt content to evolving audience preferences?▼

You can use periodic distillation to adapt content to evolving preferences by consolidating archived feedback logs over time, converting historical performance signals and edit patterns into updated, durable writing rules.

Why is my writing preference memory not capturing rejected content feedback?▼

Writing preference memory fails to capture rejected content feedback when the trigger-based learning workflow is not activated, preventing the structured logging required to parse rejections into durable rules.