Memory Flush

Promote recent logs and insights into MEMORY.md while pruning stale entries.

Updated Jun 2, 2026
One-click install
npx skills add https://github.com/Atrium-Hermes/atrium-lighthouse --skill memory-flush-atrium-hermes
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Memory Flush
Source: https://github.com/Atrium-Hermes/atrium-lighthouse/tree/main/skills/memory-flush
Command: npx skills add https://github.com/Atrium-Hermes/atrium-lighthouse --skill memory-flush-atrium-hermes

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This capability helps keep AI memory fresh by promoting high-value recent logs, lessons, and insights into MEMORY.md, ensuring durable context for future interactions.

Core Features & Use Cases

  • Selective promotion of relevant recent activity (errors, workarounds, notable findings) into memory.
  • Memory hygiene: prune stale entries and maintain concise MEMORY.md with links to details.
  • Use Case: After a sprint, promote new learnings and milestones to MEMORY.md for future reference during reviews or audits.

Quick Start

Run the memory flush operation to promote high-value memory items from recent activity into MEMORY.md.

Frequently Asked Questions about Memory Flush

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

FAQPage Schema
How do I promote recent logs and insights into long-term memory?

To promote recent logs into long-term memory, you run a memory flush operation that scans recent activity, filters for high-value learnings, deduplicates existing records, and appends selected insights to MEMORY.md.

What is the best way to maintain a concise knowledge-base and prune stale memory entries?

Maintaining a concise knowledge-base requires scheduled memory hygiene that prunes stale entries from MEMORY.md and replaces them with links to detailed logs, ensuring your long-term context remains fresh and uncluttered.

How does memory management handle deduplication when archiving recent logs?

Memory management handles deduplication by scanning existing MEMORY.md entries before archiving, filtering out redundant recent logs, and ensuring only unique, high-value insights are promoted to prevent duplicate records.

Can I selectively promote specific workarounds and errors from digests into MEMORY.md?

Yes, you can selectively promote specific workarounds and errors by applying memory flush to scan memory digests and recent logs, filtering for notable findings, and updating MEMORY.md only with selected high-value candidates.

When should I perform a memory flush on my knowledge-base system?

You should perform a memory flush on your knowledge-base system after completing significant work phases or sprints to promote new learnings, workarounds, and notable findings into MEMORY.md for future reference during audits.