Context Optimizer

Automatic context management skill that monitors token usage and creates continuation sessions when needed.

Core Functionality

This skill provides:

1. 1. Context Monitoring - Track token usage against 200k max
2. Threshold Detection - Trigger at 95% (190k tokens)
3. Information Extraction - Extract key decisions, important context, unfinished tasks
4. Session Continuation - Create new session with summarized context

Usage

Manual Trigger

CODEBLOCK0

Automatic Monitoring

Add to your workflow or cron:

CODEBLOCK1

When context exceeds threshold, the skill will:

1. 1. Analyze conversation history
2. Extract key information (decisions, context, todos)
3. Generate continuation prompt
4. Create new session with extracted context

Extraction Priority

The optimizer extracts in this order:

1. 1. Unfinished tasks - Incomplete actions, pending decisions
2. Key decisions - Important choices and their rationale
3. Critical context - Information essential to task completion
4. User preferences - Explicitly stated preferences and requirements
5. Recent progress - What has been accomplished so far

Output Format

When creating continuation, output:

CODEBLOCK2

Scripts

• - context_optimizer.py - Main CLI for context monitoring and splitting
• INLINECODE1 - Extracts key information from session history

Integration with OpenClaw

Automatic Split Logic

To automatically split sessions when context exceeds 95%, integrate with OpenClaw's heartbeat:

1. 1. Add to HEARTBEAT.md:

1. 2. Create Cron Job:

Tool Integration

The skill uses these OpenClaw tools:

• - session_status - Get current token/context usage
• INLINECODE4 - Fetch conversation for extraction
• INLINECODE5 - Create continuation session
• INLINECODE6 - Schedule periodic checks