memory-synthesizer

Classify validated work into user, repo, or session memory with JSON formatting.

Updated Apr 27, 2026
One-click install
npx skills add https://github.com/Conradmaker/copilot-cockpit --skill memory-synthesizer
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: memory-synthesizer
Source: https://github.com/Conradmaker/copilot-cockpit/tree/main/.github/skills/memory-synthesizer
Command: npx skills add https://github.com/Conradmaker/copilot-cockpit --skill memory-synthesizer

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Synthesizes durable memory signals from validated work into the correct memory scope to avoid memory pollution. This helps teams keep stable knowledge without clutter from transient details.

Core Features & Use Cases

  • Classifies outputs into user memory, repo memory, or session memory based on durability rules.
  • Applies decision rules from references/decision-rules.md to determine when to save and how to format.
  • Provides guidance if candidates should be skipped or updated, with citations to references.

Quick Start

After completing a task, provide the outputs and decisions to the memory-synthesizer so it can classify and store durable signals.

Frequently Asked Questions about memory-synthesizer

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

FAQPage Schema
How do I avoid memory pollution when storing AI session context?

To avoid memory pollution, synthesize durable memory signals by applying durability checks that classify validated work into user, repo, or session memory scopes, skipping transient notes.

When should I save outputs as repo memory versus user memory?

Apply durability decision rules to evaluate task outputs and determine whether validated work belongs in user memory, repo memory, or session memory based on long-term retention value.

How do I format repo memory entries after completing a task?

Format repo memory entries as JSON structures by following the formatting guidelines outlined in references/decision-rules.md and memories.md after evaluating validated task outputs.

How do I filter out transient notes from durable memory storage?

Apply skip rules during memory synthesis to evaluate candidate signals, filtering out transient notes and preventing non-durable details from cluttering stable knowledge storage.

What is the best way to classify task outputs into correct memory scopes?

Use a memory synthesis process implementing durability checks and skip rules to route validated task outputs into the correct memory scope, either user, repo, or session memory.