skills.rest
Wike-CHI
Wike-CHICOMMUNITY
Explore all skills by this author →
WebsiteGitHub

memory-manager

Detect memory compression and create snapshots in local agent workspaces.

2|Updated Mar 31, 2026
Search Tags:#memory#organize#compression#snapshot#semantic-memory#episodic-memory#procedural-memory
One-click install
npx skills add https://github.com/Wike-CHI/acquisition-agent --skill memory-manager-wike-chi
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-manager
Source: https://github.com/Wike-CHI/acquisition-agent/tree/main/skills/memory-manager
Command: npx skills add https://github.com/Wike-CHI/acquisition-agent --skill memory-manager-wike-chi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq.

What problem does it solve?

Local AI agents accumulate large, unstructured context that can exceed tokenizer or storage limits, causing loss of critical conversation history and operational context; this Skill detects compression risk and preserves recoverable snapshots so agents retain continuity.

Core Features & Use Cases

  • Three-tier memory architecture: organizes memories into episodic (time-based events), semantic (facts and knowledge), and procedural (workflows) stores for clearer retrieval.
  • Compression detection & alerts: estimates overall usage, warns at configurable thresholds, and surfaces recommended remediation steps.
  • Snapshots & organization tools: automatically create recovery snapshots, migrate legacy flat files into the tiered layout, and provide search and stats utilities for maintenance.
  • Use Case: Run periodic checks on a local agent workspace to detect rising context usage, snapshot recent events, and reorganize legacy notes to prevent memory loss during long-running sessions.

Quick Start

Ask the agent to run a compression check and create a snapshot for the current workspace now.

Frequently Asked Questions about memory-manager

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

FAQPage Schema
How do I prevent context loss when local agent memory reaches tokenizer limits?▼

To prevent context loss when local agent memory reaches tokenizer limits, you can run compression detection checks to monitor usage and create recovery snapshots to preserve critical conversation history and operational context.

How does a three-tier memory architecture organize episodic, semantic, and procedural memories?▼

A three-tier memory architecture organizes agent context by separating time-based events into episodic stores, facts and knowledge into semantic stores, and workflows into procedural stores, enabling clearer retrieval and maintenance.

How do I create a memory snapshot and reorganize legacy flat files for my agent workspace?▼

You create memory snapshots and reorganize legacy flat files by triggering snapshot creation utilities that migrate unstructured notes into the tiered memory layout, preventing memory loss during long-running sessions.

Do I need jq to run compression checks and memory organization scripts?▼

You do not strictly need jq to run compression checks and memory organization scripts, as the POSIX shell scripts rely on standard UNIX tools with optional jq support for enhanced JSON processing capabilities.

What is the best way to monitor local agent context usage and trigger alerts?▼

The best way to monitor local agent context usage is to use threshold-based compression detection scripts that estimate overall usage, warn at configurable limits, and surface recommended remediation steps during high-usage scenarios.

Why does unstructured agent context accumulate and exceed storage limits during long-running sessions?▼

Unstructured agent context accumulates and exceeds storage limits during long-running sessions because local AI agents continuously generate large conversation histories without automatic organization, causing critical operational context loss.

Related Skills

Explore More Software Engineering Skills→
haackedhaacked
127

analyze-permissions

Analyze Claude permission entries and propose consolidated wildcard patterns.

Community
Intermediate
binfen1binfen1

memory-manager

Organize AI agent memory into episodic, semantic, and procedural layers.

Community
Advanced
Chris-Chai-MinjaeChris-Chai-Minjae

huggingface-tokenizers

Train BPE, WordPiece, and Unigram tokenizers with Rust-based Python bindings.

Community
Intermediate
papachongpapachong
2

memory-manager

Initialize three-tier agent memory and detect compression risk with snapshots.

Community
Intermediate