Second Brain Setup Health Check
Check whether your Claude Code setup has the context, memory, skills, hooks, and quality gates it needs to do real work.
Get the Health Check
Local command, diagnosis template, and routing guide. No private repo upload needed.
Iwo Szapar
Creator of MemoryOS and Second Brain
I built this to make the health check useful as a first conversation: less audit theater, more "what is missing and what do we fix first?"
Start with the local check
Run the public health-check package in the repo where you use Claude Code. It produces a setup snapshot you can diagnose without uploading private files.
What It Checks
Setup Quality
Core files, repo shape, permission boundaries, and whether your AI can find the right context.
Usage Activity
Whether the system is actually used in live work, not just sitting there as a nice folder.
Brain Maturity
How well context, decisions, examples, and patterns compound between sessions.
Workflow Loops
Whether repeated work has been turned into skills, checklists, or agent workflows.
Quality Gates
Review, eval, and publishing checks that stop your AI from shipping sloppy work.
Next Fixes
The shortest list of improvements that will make the setup more useful this week.
How to Use It
- 1Run the health check in your Claude Code project.
- 2Copy only the summary, scores, and top recommendations.
- 3Remove secrets, customer data, repo code, and API keys.
- 4Turn the output into a setup diagnosis.
- 5Fix the first missing layer before adding more tools.
What you get back
Can the AI find goals, standards, active projects, and the current operating truth?
Are decisions, examples, patterns, and lessons saved somewhere future sessions can reuse?
Has repeated work become explicit instructions instead of repeated prompting?
Do the right checks run before a change, send, publish, or deploy?
Can recurring work move through plan, act, review, fix, and learn without drama?
Consultants
Check whether your AI remembers client context, deliverable standards, and reusable frameworks.
Founders
Find the gaps between company context, daily decisions, and the work AI is expected to handle.
Overloaded operators
Turn scattered notes, tabs, and recurring admin into a calmer setup with a real loop.
Built from the existing health check
This resource is deliberately not a new skills audit, questionnaire, or extra content layer. It points people at the current health-check package, then gives them a lightweight way to interpret the output.
The goal is simple: show the state of the setup, choose one useful fix, and route people toward the right next step.
Run the check before you add more AI stuff
Most setups do not need another tool first. They need a clearer spine: context, memory, workflow, review, and one next fix.