context-compressor

name	context-compressor
description	Compress long contexts preserving critical information. Use when use when reducing context length while keeping essential information Triggers on keywords: compression, context, optimization, tokens.
allowed-tools	Read, Write, Bash, Edit, Grep, Glob
user-invocable	true
mcp-servers	["filesystem","github","docker"]

Context Compressor

context

Compress long contexts preserving critical information. Uses hierarchical summarization, key entity extraction, and relevance scoring to fit essential context within token limits. This skill provides system-level capabilities that maintain agent state, ensure consistency, and optimize resource usage across sessions and projects.

instructions

Step 1: Assess System Needs

1. Identify the system operation required — memory, sync, compression, or role management
2. Determine scope — single session, multi-session, or cross-project
3. Map dependencies — what other systems or tools are involved
4. Establish operational parameters — retention, frequency, thresholds

Step 2: Execute System Operation

1. Initialize — Set up or load required state and configuration
2. Process — Execute the core operation with proper error handling
3. Validate — Verify the operation completed correctly
4. Report — Document what was done and any issues encountered

Step 3: Generate System Report

## System Operation Report

### Operation
Type of operation performed

### Input
What was processed

### Output
Results of the operation

### State Changes
- Change 1: before to after
- Change 2: before to after

### Issues Encountered
- Issue: resolution

Step 4: Verify System Health

• Operation completed without errors
• State is consistent and valid
• No resource leaks or orphaned data
• Logging and monitoring are active
• Backup/recovery procedures are in place

constraints

• NEVER modify system state without logging the change.
• NEVER assume operations succeed — always validate results.
• NEVER skip cleanup — system resources must be properly released.
• ALWAYS handle concurrent access and race conditions.
• ALWAYS maintain backward compatibility with existing state.

examples

Example: System Operation

Input: "Use when reducing context length while keeping essential information"

Output:

## System Operation Report

### Operation
Context Compressor executed successfully

### State Changes
- Configuration updated with new parameters
- State persisted to storage

### Verification
- All health checks passing
- No resource leaks detected