Is it safe to use during a complex refactor?

The skill includes a guide advising against compression mid-implementation to ensure the model keeps track of all recent changes until the milestone is reached.

Will I lose my code progress if I compress context?

No. Files on disk, git state, CLAUDE.md instructions, and memory files are preserved; only the intermediate conversational reasoning and tool logs are cleared.

How does this differ from Claude's built-in automatic compression?

Standard automatic compression can trigger mid-task, potentially losing important variables or file paths. Strategic Compact suggests manual compression only at logical milestones you choose.

Can I customize the frequency of compression suggestions?

Yes, you can adjust the COMPACT_THRESHOLD environment variable (defaulting to 50 tool calls) to control how often the skill prompts you.

Strategic Context Compactor

Name: Strategic Context Compactor
Author: affaan-m

byaffaan-m

•

112,919

•

생산성 및 워크플로

Optimizes Claude's performance by suggesting manual context compression at logical task boundaries rather than relying on disruptive automatic triggers.

The Strategic Compact skill is designed for high-performance development sessions where context window management is critical. It replaces arbitrary automatic compression with a guided, logic-aware approach that prompts you to run the /compact command during natural transition points, such as moving from research to planning or implementation to testing. By monitoring tool call counts and providing a decision matrix for what to retain, it ensures Claude remains coherent and responsive during complex, long-running tasks without losing vital project state or current implementation details.

주요 기능

01Integration with PreToolUse hooks for seamless, non-intrusive monitoring.

02Strategic decision matrix for transitions like Research-to-Planning or Debugging-to-Feature.

03112,919 GitHub stars

04Automatic preservation of critical state including memory files and TodoWrite lists.

05Hook-driven tool call tracking to monitor session density automatically.

06Configurable thresholds for compression suggestions based on your workflow needs.

사용 사례

01Managing massive development sessions exceeding 200K tokens while maintaining model coherence.

02Switching between unrelated project modules within the same terminal session without context pollution.

03Cleaning up extensive debugging logs and failed reasoning paths before beginning a fresh implementation phase.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code strategic-compact

For use in Claude.ai and ChatGPT

Download Skill