The agent operator’s handbook.
Calculate sandbox costs, replay a worker crash, or test a configuration with your own inputs.
Start with a decision.

Sandbox cost crossover
For a model-bound worker, CPU utilization can change which sandbox costs less. Compare identical memory capacity and see the utilization where the two compute bills cross.

Crash and replay simulator
Move the crash across a real job timeline. See which steps remain durable, which steps replay, and which external writes need reconciliation.

First-run checkpoint checker
Check Node.js against OpenClaw’s documented version ranges, then record separate times for opening the dashboard and receiving a fresh model reply. Keep those observations with the active config path so you can reproduce a failure.
Find a field guide.
Search 232 guides by platform, workflow, or decision.
- AI Agent Hosting for Agency Handoffs
Client workspace isolation audit
- AI Agent Hosting Under Five Dollars
Five-dollar hosting envelope
- Benchmark Claude API Models Before Shipping
Paired model trial report
- Build a Claude Skill Buyers Can Accept
Skill output acceptance manifest
- Build a No-Code Document Agent
Document extraction field audit
- Build an OpenClaw SaaS with Safe Retries
Retry receipt reconciliation
- Can a Claude Agent SDK Worker Survive Replacement?
Worker Replacement Drill
- Can a Fresh Worker Resume Your Claude Agent?
Worker Resume Gate
- Can a Managed Agents Worker Pass Your Canary?
Sandbox canary verifier
- Can a Timer Process Booking Requests Before Cutoff?
Calendar booking replay
- Can a white-label OpenClaw client leave cleanly?
Client exit receipt checker
- Can batch review keep up with automation exceptions?
Exception queue replay
- Can Claude Code Finish Before Your Allowance Resets?
Claude allowance checkpoint planner
- Can Codex Cloud Pass Your Worker Acceptance Test?
Worker Acceptance Table
- Can Deep Agents Resume Work After Sandbox Loss?
Sandbox replay audit
- Can Firebase Events Survive a Reconnect?
HN Refresh Replay
- Can Hermes Agent or Claude Code Clear a Backlog?
Backlog recovery calculator
- Can Meta Robyn Replace Google Meridian for Budget Decisions?
Lift Prediction Check
- Can Only Approved Users Reach OpenClaw Web Chat?
Tailnet access matrix
- Can OpenClaw Intake Feed Claude Code Without Duplicate Work?
Request Handoff Reconciler
- Can QM Keep Team Decisions in the Right Scope?
Decision Scope Audit
- Can the App and Worker Survive Independently?
App and Worker Receipt Replay
- Can This Docker Sandbox Move to Cloud?
Docker move manifest audit
- Can This Mac Mini Run OpenClaw Unattended?
Mac Mini Operating Budget
- Can You Administer Hermes Without Exposing Its Dashboard?
Dashboard Reachability Audit
- Can Your Agent Leave Modal Without Double-Submitting Jobs?
Paired Replay Audit
- Can Your Browser Console Trace a Stalled Run?
Console Checkpoint Audit
- Can Your Claude Code Channel Handle a Restart?
Channel probe audit
- Can Your Daytona Deployment Drain an Outage Backlog?
Daytona Backlog Drill
- Can Your EU Hermes Gateway Meet Recovery Targets?
Recovery Drill Ledger
- Can Your Free Automation Survive a Retry?
Queue replay and quota test
- Can Your GitHub Actions Gate Reject a Slow Agent?
Paired Timing Gate
- Can Your MCP Server Meet Its Client Deadline?
MCP route comparison
- Can Your OpenAI Agents SDK Job Resume Safely?
Sandbox Recovery Log Check
- Can Your Sandbox Report Survive Session Expiry?
Sandbox report acceptance check
- Can Your Self-Hosted Sandbox Survive a Worker Loss?
Worker Loss Burst Replay
- Can Your Voice Setup Preserve Coding Instructions?
Dictation Trial Audit
- Change OpenClaw’s Default Model Safely
Model cutover evidence timeline
- Choose a Manus AI Replacement
Research brief coverage comparison
- Choose a No-Code Agent Builder
Builder acceptance matrix
- Choose OpenClaw Cloud Hosting by Recovery
Recovery failure budget comparison
- Choose Your First OpenClaw Messaging Channel
Messaging channel delivery trial
- Claude Code or Custom GPT?
Platform handoff comparison
- Claude Code Servers That Survive Replacement
Transcript and workspace restore diff
- Compare LLM Prices per Accepted Job
Cost per accepted inference
- Configure OpenClaw’s Slack Transport
Slack transport handoff tracer
- Configure Telegram Access in OpenClaw
Telegram access-policy diff
- Create a Claude Skill That Triggers
Skill trigger precision and recall
- Create a Telegram Bot with BotFather
Bot registration handoff receipt
- Diagnose a Silent OpenClaw Telegram Bot
Telegram silence trace reader
- Did OpenClaw Use the Provider You Configured?
Provider route evidence checker
- Distribute AI Agents Through Buyer Workflows
Channel activation funnel
- Do you need OpenClaw, Codex Cloud, or both?
Chat-to-Code Trace Audit
- Does Claude Code Need a VM for This Workflow?
Claude sandbox fixture audit
- Does Your Claude Code Workload Justify a Subscription?
Claude Code Billing Crossover
- Does Your Coding Agent Upgrade Earn Its Cost?
Paired Benchmark Break-Even
- Does Your Support Queue Need an Agent?
Matched support replay
- Find Where OpenClaw Startup Stalled
Gateway startup trace comparison
- Get and Test an OpenAI Key
API credential acceptance log
- How Do You Commission a Text-First AI Assistant?
Inbox Acceptance Check
- How Do You Verify a Hermes Agent Uninstall?
Hermes removal verifier
- How Do You Verify DeepSeek in OpenClaw?
DeepSeek paired fixture check
- How Do You Verify Grok API-Key Routing?
Grok cutover audit
- How Do You Verify the Hermes Agent API?
Hermes API probe checker
- How Fast Can Background Agents Clear a Backlog?
Backlog Drain Simulator
- How Many Kubernetes Sandboxes Should Stay Warm?
Warm Pool Burst Replay
- How Many OpenClaw Jobs Can Run for Free?
Search Quota Planner
- How Many Parallel Agents Can Review Keep Up With?
Agent review queue replay
- How Much OpenClaw Work Can You Recover?
Restore and Replay Budget
- How Much RAM Does Your Claude Code VPS Need?
VPS memory envelope
- How Much Retry Traffic Can Your Agent Safely Add?
Retry Capacity and Deadline
- How much work survives a worker crash?
Crash and replay simulator
- Install and Update OpenClaw on Ubuntu
Release and rollback file comparison
- Is Hermes Ready for an Unattended Gateway?
Gateway Probe Ledger
- Is your OpenClaw deployment ready for traffic?
Release sign-off matrix
- Managed OpenClaw Hosting That Survives Updates
Managed update continuity audit
- MCP or Claude Skills for Approvals?
Approval trace checker
- Migrate a scheduled Claude Cowork workflow
Scheduled workflow cutover reconciliation
- Monetize Claude Code Skills by Acceptance
Skill offer break-even worksheet
- Move a Runpod workload without losing state
GPU cutover cost crossover
- OpenClaw Hosting for Agency On-Call
Agency on-call response budget
- Pinning Claude Skills API Releases
Skill release manifest diff
- Price AI Agents for Peak Usage
Peak usage margin calculator
- Price OpenClaw Skills by Accepted Outcome
Skill support allowance planner
- Qualify Local AI Models Under Load
Local inference capacity trial
- Run Claude Code Overnight
Overnight completion deadline
- Running Claude Skills Without Desktop
Headless output manifest checker
- Sell AI Workflows by Accepted Deliverable
Deliverable invoice reconciliation
- Set Up OpenClaw Discord Access
Discord access-path comparator
- Share Claude Skills by Update Owner
Skill rollout drift report
- Should Modal Compute and Agent State Run Separately?
GPU Checkpoint Break-Even
- Should OpenCode Run Per Job or Stay Available?
OpenCode Worker Cost Crossover
- Should Telegram Intake Share Your OpenCode Runner?
Chat-to-diff receipt audit
- Should You Buy or Publish a Claude Skill?
Skill Fixture Audit
- Verify Kimi K2.5 in OpenClaw
Kimi provider-route checkpoint test
- Verify OpenClaw’s OpenAI API Key
OpenAI billing-route trial log
- Verify Your Claude Key in OpenClaw
Anthropic authentication trial log
- What Changes When You Move to Daytona V2?
Sandbox Cost Crossover
- What does idle sandbox time cost?
Sandbox cost crossover
- What Should a Managed OpenClaw Contract Cover?
Managed OpenClaw Quote Audit
- What Should OpenClaw Managed Hosting Restore?
OpenClaw restore drill
- What Will Claude Managed Agents Cost Per Accepted Job?
Managed Agents Batch Ledger
- What will your E2B workload cost each month?
E2B burst cost replay
- When Do AgentCore Runtime Instances Cost Less?
AgentCore cost crossover
- When Does a Larger Modal Sandbox Pay Off?
Modal Sandbox replay ledger
- When Does a MicroVM Cost Less per Job?
Runtime Cost Crossover
- When Does Claude Code Max Beat Pro Plus API?
Subscription overflow replay
- When Does Cloudflare Cost Less Than Vercel?
Sandbox metering crossover
- When Does Codex Cloud Need a Persistent Host?
Host Backlog Replay
- When Does Daytona Cost Less Than Vercel Sandbox?
Sandbox Meter Comparison
- When Does E2B or Daytona Cost Less?
Sandbox retention cost crossover
- When Does E2B or Modal Cost Less to Resume?
Sandbox Return Cost Comparator
- When Does gVisor Meet Your Hosted Agent Requirements?
Sandbox Trial Comparator
- When Does Managed OpenClaw Hosting Beat a VPS?
Hosting Labor Break-Even
- When Does Metered Compute Lower OpenClaw Costs?
Worker Cost Crossover
- When Does Metered Hosting Beat a Mac Mini?
Metered Hosting Crossover
- When Does OpenClaw Beat Claude Cowork for Intake?
Intake deadline replay
- When Does Running Hermes Cost Nothing Extra?
Hermes monthly cost ledger
- When Is a Dedicated Hermes VPS Worth It?
VPS Deadline Break-Even
- When Is an AI Agent Ready for Marketplace Sales?
Activation capacity planner
- When is an OpenClaw gateway ready for traffic?
OpenClaw release gate
- When Is Disposable Claude Code Hosting Cheaper?
Hosting cost crossover
- When Is E2B Embed Cheaper Than Managed Compute?
E2B Monthly Crossover
- When Is Leaving Hostinger Worth the Move?
Host Migration Break-Even
- When Is Managed Hermes Hosting Cheaper Than Hetzner?
Hermes hosting break-even
- When Is Sandbox Memory Worth Paying For?
Sandbox replay break-even
- When Is Self-Hosting Cursor Cloud Agents Worth It?
Cursor Hosting Break-Even
- When Is Waiting for Cheaper Hetzner Stock Worthwhile?
Stock wait break-even
- When OpenClaw Stops Answering on WhatsApp
WhatsApp delivery receipt timeline
- When Should Hermes Escalate to a Stronger Model?
Model Escalation Budget
- When Should OpenClaw Escalate Beyond DeepSeek Flash?
Fallback Replay Ledger
- When should you choose Claude Code for an agent task?
Agent trial comparator
- When Should You Keep a Claude Code Host?
Claude Code handoff break-even
- When to Use Autonomous AI Agents
Autonomy stop-rule planner
- Where Can Claude Code Worktrees Still Collide?
Worktree collision check
- Where Can OpenClaw Meet Your Availability Target?
Host Deadline Replay
- Where Did Your Agent Stop Making Progress?
Agent Trace Timer
- Where Did Your OpenClaw First Run Stall?
First-run checkpoint checker
- Where does your OpenClaw iMessage delivery fail?
iMessage probe trace checker
- Where should each OpenClaw Gateway run?
Gateway inventory audit
- Where Should You Run Claude Code Unattended?
Unattended Reply Trial
- Where Should Your Hermes Gateway Run?
Hermes Host Schedule Replay
- Where to Self-Host Claude Code
Self-host capacity and reserve check
- Which AEO Agency Can Show Citation Lift?
Paired Citation Pilot
- Which AEO Tool Finds Citations You Can Verify?
Citation Disagreement Audit
- Which Agent Boundary Failed Your Acceptance Test?
Boundary Evidence Checker
- Which Agent Builder Survives a Replayed Request?
Replay Receipt Auditor
- Which Agent Can Clear Your Overnight Queue?
Overnight Backlog Projection
- Which Agent Can Share Your Team’s Workflow?
Agent pilot audit
- Which Agent Components Must Stay on Your Infrastructure?
Runtime Placement Audit
- Which Agent Handles Your Workload With Less Intervention?
Agent Workload Trial
- Which Agent Host Can Clear Your Outage Backlog?
Hosting recovery calculator
- Which Agent Host Clears a Backlog Within Your Deadline?
Backlog replay comparator
- Which Agent Layer Fixes Your Workflow Failure?
Agent Workflow Fixture Diff
- Which agent platform clears your outage backlog?
Replay Capacity Calculator
- Which Agent Runs Produced Verifiable Results?
Run Receipt Audit
- Which Agent Steps Can Your Trace Actually Prove?
Trace Coverage Audit
- Which agent survives a procedure revision?
Revision replay checker
- Which Agent Test Platform Catches Your Release Regressions?
Paired Verdict Replay
- Which AI Platform Fits Your Nonprofit’s First Workflow?
Nonprofit Pilot Workload Audit
- Which AI Tracker Preserves Evidence You Can Audit?
Citation export audit
- Which Android Emulator Setup Delivers On-Time Passes Cheapest?
Emulator pilot ledger
- Which assistant can you pilot without an invite?
Follow-up access audit
- Which Claude Code alternative fits a zero-charge budget?
Zero-charge coding capacity
- Which Claude Code Alternative Survives Your Runner Constraints?
Coding runner dependency audit
- Which Claude Code Changes Cut Cost per Accepted Fix?
Accepted fix cost ledger
- Which Claude Code remote setup survives a handoff?
Handoff drill comparison
- Which Claude Code Server Survives Your Outages?
Server recovery drill
- Which Claude Cowork Alternative Fits Your Review Queue?
Report Review Capacity
- Which Claude Skills Are Worth Hosting?
Hosted skill pilot budget
- Which Claude Slack Setup Meets Your Delivery Contract?
Slack Delivery Trace Audit
- Which Cloudflare Container Timeout Costs Less on Replay?
Container Timeout Replay
- Which Codex Alternative Produces Reviewable Patches on Time?
Patch Handoff Benchmark
- Which Coding Agent Costs Less per Accepted Change?
Coding Trial Ledger
- Which Coding Agent Fits Your Actual Job?
Coding Agent Job Matrix
- Which Coding Agent Setup Shortens the Release Path?
Release Schedule Simulator
- Which Coding Agent Stack Recovers Without Duplicate Work?
Handoff Recovery Comparison
- Which Coding Agent Survives a Phone App Switch?
Mobile Resume Audit
- Which Daytona Alternative Passes Your Portability Test?
Cold Run Portability Check
- Which Docker Host Passes Your Agent Recovery Drill?
Docker recovery and replay budget
- Which fleet sandbox migrations can you approve?
Fleet Destination Audit
- Which free OpenClaw host recovers fast enough?
Host replay comparison
- Which Gateway Is Handling Your WhatsApp Messages?
WhatsApp Handoff Trace
- Which Guardrail Actually Stopped the Agent?
Guardrail Receipt Audit
- Which Handoffs Belong in Cowork or OpenClaw?
Handoff trial audit
- Which Hermes Deployment Reaches Readiness Fastest?
Hermes readiness clock
- Which Hermes Docker upgrade is safe to promote?
Hermes mount comparison
- Which Hermes host can clear your outage backlog?
Hermes backlog comparison
- Which Hermes Install Passes Your Upgrade Rehearsal?
Hermes Upgrade Rehearsal
- Which Hermes Interface Fits Your Operator Tasks?
Hermes Operator Trial
- Which Hermes Interface Reaches the Right Profile?
Hermes profile audit
- Which Hermes Replacement Preserves Your Workflows?
Workflow Parity Ledger
- Which Hermes skill revision passes your release tests?
Skill Revision Diff
- Which Hetzner VPS Sustains Your Hermes Workload?
VPS Replay Comparator
- Which LLM Deployment Passes Your Data Residency Rules?
Residency Manifest Check
- Which Local Build Passes Your OpenClaw Workload?
Local build trial comparator
- Which Managed OpenClaw Host Can You Exit?
Managed Exit Drill
- Which OpenClaw Alternative Fits Your Operating Constraints?
Migration Acceptance Matrix
- Which OpenClaw Cloud Setup Clears Downtime Backlogs?
Cloud Backlog Rehearsal
- Which OpenClaw Host Passes Your Peak Workload?
Peak Burst Comparison
- Which OpenClaw Jobs Belong in Runtime Hooks?
Runtime rehearsal check
- Which OpenClaw VPS Can Drain Your Outage Backlog?
VPS Catch-Up Planner
- Which OpenClaw VPS Passes Your Capacity Trial?
VPS Capacity Trial
- Which OpenClaw Workflows Survive a Replay?
Reconciliation Replay Audit
- Which Part of Claude Managed Agents Should You Replace?
Runtime Recovery Ledger
- Which Personal Assistant Preserves the Right Reply Identity?
Assistant Reply Audit
- Which Prompt Library Adds Evidence You Can Publish?
Evidence Retention Audit
- Which QM Workflows Need a Shared Workspace?
QM Migration Trial
- Which Railway Alternative Clears Your Job Burst?
Sandbox Burst Replay
- Which Remote Runtime Can Build Your Checkout?
Remote Runtime Preflight
- Which Repository Boundaries Does Your Agent Actually Respect?
Repository Boundary Audit
- Which Review Agent Pays Back Maintainer Time?
Review Triage Break-Even
- Which Sandbox Can Recover Before Your Deadline?
Sandbox Recovery Budget
- Which Sandbox Recovers Within Your Deadline?
Sandbox recovery drill
- Which Vercel Sandbox Alternative Meets Your Recovery Target?
Sandbox recovery comparison
- Which VPS Can Handle Your Agent’s Busiest Job?
VPS Burst Worksheet
- Which VPS fits your Claude Code workload?
VPS fit and quote comparator
- Which VPS Handles Your OpenClaw Peak?
VPS Peak Memory Envelope
- Which VPS Passes Your Claude Code Load Test?
VPS replay comparison
- Which white-label agent design can your agency support?
Client pilot staffing comparison
- White-Label Hermes Hosting With Portable Skills
Portable skill package audit
- Who Keeps Your Agent Reachable After Hours?
Channel Outage Replay
- Will a Kata Pool Fit Through Maintenance?
Kata Drain Capacity
- Will Claude Code Channels Recover After a Restart?
Channel restart ledger
- Will Codex State Survive Your Docker Image Upgrade?
Docker mount drift audit
- Will Hermes Compose Reattach the Right State?
Compose Mount Diff
- Will Hermes Compose Reuse the Setup State?
Hermes Bind Mount Check
- Will OpenClaw Compose preserve the right paths?
Compose mount coverage check
- Will OpenClaw or Zo Computer Pass Your Channel Trial?
Channel Trial Comparator
- Will Your Agent Stall Alert Fire Too Early?
Milestone Alert Replay
- Will Your Cursor Worker Pool Meet Start Deadlines?
Cursor pool burst replay
- Would Switching Voice Carriers Recover More Orders?
Voice order replay