Scan Teardown
$25
- Run the free scan
- Send redacted output after checkout
- Same-day prioritized risk notes
- Top fixes for memory, restore, launchd, and outbound automation
Run local AI agents without giving them the keys to everything
A same-day Mac local-agent risk teardown for Codex/OpenClaw users whose memory, tmux restore, launchd jobs, or outbound automation are starting to feel unsafe.
Buy the $25 scan teardown Buy the $49 kit View sample Read guide
Buy The Teardown, Kit, Or Setup
Safe Agent Ops Kit
$49
- Approval policy
- Tenant boundary checklist
- Skill review worksheet
- launchd audit template
- Weekly operator review
Mac Agent Diagnostic
$99
- Same-day risk snapshot
- Tool, skill, and persistence review
- Top five hardening fixes
- Recommended setup path
Managed Setup
$499
- Private workspace layout
- Approval defaults and deny lists
- launchd persistence cleanup
- Backup and rollback runbook
- One safe automation path defined
Who Needs This
If any of these are true, buy the kit before adding more agent sessions, skills, browser profiles, or launchd jobs.
- Your agent can send email, post publicly, or message leads.
- Your terminal sessions auto-restore and consume memory after reboot.
- You have unreviewed skills, MCP servers, or shell tools installed.
- You work across multiple client or project workspaces on one Mac.
- You run agents with broad filesystem access.
- You do not have a written approval policy for outbound actions.
Fast path
Buy the $25 teardown if you want a same-day read on your scan output. Buy the $49 kit if you can implement checklists yourself. Buy the $99 diagnostic if you want a deeper prioritized review. Buy setup if you want the hardened operating model installed for you.
Free Guide
Runaway memory, automatic session restore, and broad agent permissions are the same class of problem: local agents need operating boundaries before they touch real work.
Read the Codex/OpenClaw runaway-agent safety guide.
Free read-only scan
Use the scan to check process memory, tmux sessions, custom launch agents, Mail Outbox queue size, and OpenClaw/Codex state size before buying.
Download the Mac agent risk scan.
After buying the $25 teardown, send redacted scan output to igor@openclaw.ai with your Stripe receipt email.
Teardown Intake
Do not send secrets, API keys, private customer data, or full file contents. Send the scan output plus one paragraph describing what went wrong.
Why This Exists
OpenClaw-style local agents are useful because they can interact with files, shell, browser sessions, network resources, and business tools. That same surface is dangerous when a business owner installs untrusted skills or runs broad automation without approval gates.
Positioning: this is not a generic prompt pack or raw gateway rental. It is a boring hardening package: minimal permissions, known tools, human approval, repeatable rollback, and clear operator handoff.
5core templates
$25scan teardown
$99diagnostic upsell
Same daymanual fulfillment
View the public sample. The full editable ZIP is delivered after checkout.
Delivery
After checkout, the full editable ZIP is delivered same-day to the email collected by Stripe. The paid archive is stored in private fulfillment storage, not on the public GitHub Pages site.
For scan teardowns, send redacted scan output to igor@openclaw.ai after checkout.
The public repo contains the sales page and sample only. The paid archive is private.
Included Controls
- Dedicated workspace layout for clients, workflows, logs, exports, and backup notes
- Explicit disallow list for sensitive accounts, folders, and destructive actions
- Dry-run mode for outreach and production writes
- Human approval before email, LinkedIn messages, payments, refunds, public posts, deletes, and production changes
- Review of existing launch agents, daemons, and listening ports