StartClaw-Optimizer

Master optimization system - APPLIES TO EVERY RESPONSE. Before responding, classify task complexity (simple question vs analysis vs coding). Use Haiku for simple/navigation/extraction/status. Use Sonnet ONLY for writing/analysis/planning/debugging. Monitor context size - if >50k tokens, recommend /compact. For automations, use scheduler wrapper. Never load full conversation history for simple tasks. Heartbeats always Haiku, single-line only. Never use Opus. This skill MUST run before every response to prevent 100k+ token bloat.

Instalar
$clawhub install startclaw-optimizer

OpenClaw Optimizer Skill

Overview

The OpenClaw Optimizer is a comprehensive performance and cost optimization skill designed to enhance the efficiency of Clawdbot's subagent workflows.

Core Components

1. Task Router

  • Intelligent model selection based on task complexity

  • Automatic routing between Haiku, Sonnet, and Opus models

  • Cost prediction and optimization

2. Scheduler

  • Robust task execution with retry mechanism

  • Configurable preflight and postflight hooks

  • Timeout and exponential backoff handling

3. Browser Governor

  • Browser tab serialization

  • Concurrent tab management

  • Circuit breaker for preventing runaway processes

4. Context Compaction

  • Automatic token tracking

  • Summarization at 50,000 token threshold

  • Preserves critical context

5. Real-time Dashboard

  • Monitor daily budget

  • Track task execution

  • Visualize model distribution

  • Circuit breaker status

Expected Savings

Before Optimization: $90/day After Optimization: $3-5/day Savings: 95%

Installation

npm install @startclaw/openclaw-optimizer

Usage

const { TaskRouter, OptimizerScheduler, BrowserGovernor } = require('@startclaw/openclaw-optimizer');

const router = new TaskRouter();
const scheduler = new OptimizerScheduler();
const browserGovernor = new BrowserGovernor();

// Automatic model and cost optimization
const modelSelection = router.selectModel(taskDescription);
await scheduler.execute(task, modelSelection);

Monitoring


# Real-time dashboard
python3 scripts/dashboard.py watch

License

StartClaw Internal Use License