strategic-compact

Prompt manual /compact at logical workflow boundaries using configurable thresholds.

Updated Jan 30, 2026
One-click install
npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill strategic-compact-cobaltsato
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: strategic-compact
Source: https://github.com/CobaltSato/avalanche-build-games-tool-kit/tree/main/.claude/skills/strategic-compact
Command: npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill strategic-compact-cobaltsato

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Strategic context can be lost when auto-compaction is triggered at arbitrary points. This skill prompts manual /compact at logical boundaries to preserve relevant information for upcoming phases.

Core Features & Use Cases

  • Tracks tool usage during a session to identify when context is still fresh or stale.
  • Applies configurable thresholds and hook-based prompts to trigger user-initiated compaction.
  • Use Case: During long experiments, compact after exploration or after completing a milestone to reset context for the next phase.

Quick Start

Install the PreToolUse hook and set COMPACT_THRESHOLD to your preferred limit.

Frequently Asked Questions about strategic-compact

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

FAQPage Schema
How do I prevent losing strategic context during automatic compaction in long-running tasks?

Manual context compaction preserves strategic information by prompting /compact at logical workflow boundaries. Using PreToolUse hooks with a configurable threshold, it triggers compaction after milestones like exploration or debugging instead of arbitrary points.

What is a PreToolUse hook for context management and how does it work?

A PreToolUse hook intercepts tool calls to track session usage against a configurable COMPACT_THRESHOLD. When tool calls exceed the limit, it prompts the user to manually initiate context compaction at a logical boundary.

When should I compact context in sessions that span exploration, planning, and implementation?

You should compact context after completing exploration or hitting a milestone to reset for the next phase. Prompting manual compaction at these logical boundaries ensures relevant information is preserved for upcoming implementation and debugging tasks.

Can I configure the tool-call threshold that triggers manual compaction prompts?

Yes, you can set the COMPACT_THRESHOLD variable to your preferred tool-call limit. Once the session counter reaches this threshold, the hook prompts you to manually execute /compact, ensuring context management aligns with your workflow.

Does manual context compaction work better than auto-compaction for long experiments?

Manual context compaction is better for long experiments because it targets logical workflow boundaries rather than arbitrary trigger points. This prevents the loss of strategic context needed for subsequent planning, implementation, and debugging phases.