lcmp_recommendation

Analyze recent work context and categorize durable items into LCMP files.

2|Updated Nov 15, 2025
One-click install
npx skills add https://github.com/grandinh/claude-chaos-express --skill lcmp-recommendation
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: lcmp_recommendation
Source: https://github.com/grandinh/claude-chaos-express/tree/main/.claude/skills/lcmp_recommendation
Command: npx skills add https://github.com/grandinh/claude-chaos-express --skill lcmp-recommendation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Over time, conversational context can become bloated with ephemeral details, making it harder to retrieve durable insights. This Skill helps you identify and promote valuable information to long-term memory (LCMP files), ensuring knowledge is preserved.

Core Features & Use Cases

  • Compaction Opportunity Analysis: Reviews recent work context to identify durable information (decisions, insights, gotchas) worth preserving in LCMP files.
  • LCMP Categorization: Suggests where to promote information (e.g., decisions.md, insights.md, gotchas.md) with proposed formats and rationale.
  • User-Driven Compaction: Provides recommendations for context compaction, but never auto-modifies files without explicit user approval, giving you full control.
  • Use Case: After completing a major feature, you can use /squish to get recommendations on what architectural decisions, new patterns, or critical gotchas should be permanently recorded in your LCMP documentation.

Quick Start

Squish the key learnings from our recent authentication feature into the LCMP files.

Frequently Asked Questions about lcmp_recommendation

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

FAQPage Schema
How do I preserve key decisions and insights from my work context before it gets too large?

Context compaction identifies durable information—architectural decisions, patterns, and critical gotchas—worth promoting to long-term memory files like decisions.md, insights.md, and gotchas.md. Use the `/squish` command after completing major work to surface what belongs in permanent documentation rather than ephemeral chat history.

What information should I move to decisions.md, insights.md, and gotchas.md files?

Decisions.md captures architectural choices and rationale. Insights.md holds patterns, techniques, and learnings discovered during implementation. Gotchas.md documents edge cases, pitfalls, and constraints encountered. The compaction analysis reviews recent context and categorizes findings into these three files with proposed formats and rationale.

When should I use context compaction instead of letting conversation history grow?

Use compaction after completing a major feature, debugging session, or architectural exploration when you've accumulated valuable learnings worth retaining. Compaction prevents bloated context from making it harder to retrieve durable insights and ensures knowledge survives across future conversations.

Does context compaction automatically modify my files?

No. Compaction provides recommendations and suggested formats for what to preserve, but requires your explicit approval before any edits. You retain full control over what gets promoted to long-term memory and how it's documented.

What's the difference between ephemeral context details and durable information worth preserving?

Ephemeral details are specific command outputs, debugging steps, or trial-and-error attempts. Durable information includes decisions and their rationale, reusable patterns, architectural tradeoffs, and critical constraints or gotchas. Compaction identifies which recent work contains lasting value for future reference.