skills.rest
CC90210
CC90210COMMUNITY
Explore all skills by this author →
WebsiteGitHub

memory-compression

Store cross-session tool usage and file edits in SQLite with semantic search.

1|Updated Apr 18, 2026
Search Tags:#automation#context#plugin#memory#sqlite#claude-mem
One-click install
npx skills add https://github.com/CC90210/CMO-Agent --skill memory-compression
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-compression
Source: https://github.com/CC90210/CMO-Agent/tree/main/skills/memory-compression
Command: npx skills add https://github.com/CC90210/CMO-Agent --skill memory-compression

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bun, claude-mem.

What problem does it solve?

This skill solves the issue of context loss between AI sessions by automatically compressing and indexing tool usage, file edits, and session observations into a persistent SQLite database.

Core Features & Use Cases

  • Automatic Memory Persistence: Captures and summarizes tool activity across different sessions without manual intervention.
  • Semantic Search: Enables the AI to retrieve past decisions, file changes, and project history using vector embeddings.
  • Use Case: When starting a new session, use this skill to quickly recall the specific technical decisions or command-line arguments used in a project three weeks ago.

Quick Start

Trigger the memory compression worker by starting your session with the command startup to initialize the database and begin tracking your current workflow.

Frequently Asked Questions about memory-compression

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

FAQPage Schema
How do I persist Claude Code context across different AI sessions?▼

Cross-session memory is maintained by automatically compressing tool usage and file edits into a persistent SQLite database, allowing Claude Code to retain historical context without manual intervention.

How does semantic vector search retrieve past project decisions?▼

Semantic vector search retrieves past project decisions by indexing session observations and file modifications into embeddings, enabling the AI to query specific technical decisions from previous workflows.

Do I need Bun and claude-mem to enable automatic memory compression?▼

Yes, enabling automatic memory compression requires the Bun runtime and the claude-mem plugin to initialize the SQLite database and maintain persistent observation logs across sessions.

What is the best way to recall file changes from a Claude Code session weeks ago?▼

Recalling file changes from weeks ago is best handled by querying the indexed historical context, which automatically captures and summarizes file modifications and command-line arguments for later retrieval.

How do I initialize the SQLite database for tracking project activity?▼

To initialize the SQLite database for tracking project activity, trigger the memory compression worker by starting your session with the startup command to begin logging your current workflow.

Related Skills

Explore More Software Engineering Skills→
vinhnxvvinhnxv
6

cc-inspect

Inspect Claude Code runtime state and plugin configuration via Bash script.

Community
Intermediate
JPFrancoiaJPFrancoia

codecompanion-memory

Retrieve past CodeCompanion conversations from a local ChromaDB via semantic search.

Community
Intermediate
kivo360kivo360
1

moai-cc-memory-template

Define a standard session summary format for memory entries.

Community
Basic
bmsull560bmsull560

memory-context

Integrate vector-based memory with knowledge graphs for multi-session conversation context.

Community
Advanced