Openclaw Agent Optimize

# Openclaw Agent Optimize Optimize OpenClaw workspaces with structured audits, cost-aware model routing, and context de-bloating for leaner, more reliable AI agent operations. Openclaw Agent Optimize is a diagnostic and strategic tool designed to improve the efficiency of your AI agent environment. It performs structured audits to identify high-cost model routing, context bloat, and inefficient delegation patterns. Instead of making silent changes, it provides a prioritized plan with exact change proposals and rollback steps to ensure your workspace remains stable and cost-effective. ## How It Works - **advisory-first**: model. It analyzes your workspace configuration and session history to produce an executive summary of top drivers affecting cost and reliability. It then presents - **Options A/B/C**: with specific tradeoffs. Once you approve a plan, it generates the exact patch proposal and verification steps. To ensure safety, it requires a new session to pick up snapshot changes and always provides a proven rollback path. ## Key Features - **Cost-Aware Routing**: Proposes specific model routing for engineering, notifications, and research tasks to minimize spend. - **Context Discipline**: Identifies and trims "transcript noise" and oversized bootstrap files that slow down responses. - **Specialist Surface Reduction**: Moves low-frequency skills to on-demand subagents to keep the main prompt surface lean. - **Safety Contract**: Guarantees no persistent mutations or cron job updates occur without explicit user approval. ## Requirements - **LLM API Key**: Required to perform workspace analysis and generate optimization proposals. Supports: OpenAI, Anthropic, Google Gemini. ## Use Cases - **Reducing Operational Costs**: Identify which tools or crons are driving up token usage and implement silent success loops. - **Fixing Context Bloat**: Clean up slow replies by identifying top offenders in your skills and bootstrap files. - **Improving Reliability**: Audit your delegation posture to ensure reasoning-heavy tasks are routed to the appropriate models. ## Installation Install via: `npx clawhub@latest install openclaw-agent-optimize`

Installation

Compétences associées