GitHub - bernardohcrocha/claude-code-os: The operational co-pilot that self-updates. Self-improving, proactive, local-first. · GitHub
/" data-turbo-transient="true" />
Skip to content
Search or jump to...
Search code, repositories, users, issues, pull requests...
-->
Search
Clear
Search syntax tips
Provide feedback
--><br>We read every piece of feedback, and take your input very seriously.
Include my email address so I can be contacted
Cancel
Submit feedback
Saved searches
Use saved searches to filter your results more quickly
-->
Name
Query
To see all available qualifiers, see our documentation.
Cancel
Create saved search
Sign in
/;ref_cta:Sign up;ref_loc:header logged out"}"<br>Sign up
Appearance settings
Resetting focus
You signed in with another tab or window. Reload to refresh your session.<br>You signed out in another tab or window. Reload to refresh your session.<br>You switched accounts on another tab or window. Reload to refresh your session.
Dismiss alert
{{ message }}
bernardohcrocha
claude-code-os
Public
Notifications<br>You must be signed in to change notification settings
Fork
Star
main
BranchesTags
Go to file
CodeOpen more actions menu
Folders and files<br>NameNameLast commit message<br>Last commit date<br>Latest commit
History<br>42 Commits<br>42 Commits
scripts
scripts
templates
templates
ARCHITECTURE.md
ARCHITECTURE.md
FRAMEWORK.html
FRAMEWORK.html
LICENSE
LICENSE
README.md
README.md
claude-code-os.png
claude-code-os.png
setup.sh
setup.sh
View all files
Repository files navigation
Claude Code OS
The Operational Copilot that Self-Updates.
Self-evolving · No extra tools · No workflow changes · Token-efficient · Auto git backup
Claude Code executes well. The problem is it has no persistent operational memory .
Every session starts from zero. Every new feature, every new document, every detail that changed in your project: you had to re-explain it, or dump the entire project as context and pay the token cost. Either way, you're doing work the agent should be doing.
Claude Code OS fixes that. A _brain/ folder lives inside your project root. It reads from your code, docs, and configs on first run. From then on, it updates itself every day, automatically, based on what actually changed.
It already knows.
You ship a new payment method. It updates the product knowledge, the customer segments it unlocks, and the metrics to start tracking. You never said a word.
You add brand assets to a folder. Brand guidelines updated automatically.
You come back after a month on a new machine. It reads the files and continues without skipping a beat. No re-explaining your stack. No copy-pasting context. Nothing lost.
It acts. Not just remembers.
Ask questions no simple dashboard can answer. Or schedule tasks that run automatically, with full context already loaded.
→ "Which customers dropped usage 30%+ this month? Cross-check their support history and draft a personalized re-engagement message for each."
→ "Every Monday: pull last week's numbers from Stripe, compare against the previous week, flag anomalies, and queue a follow-up for any account that dropped below 10% quota."
→ "Find signups from the last 30 days with no activity after signup. Filter out fake-looking domains. Visit each company's website, understand what they do, map it to how similar customers already use the product, and write a personalized email for each, leading with the pain points most relevant to their specific use case."
Set once. Runs automatically. Already has full context because it lives in the project.
Your live command center.
Open _brain/dashboard.html in your browser. It auto-refreshes every 5 minutes.
No fixed template. The dashboard is built from your actual data and evolves as your business does. Different sections, different metrics, different priorities based on what matters right now.
Scheduled tasks, proactive suggestions, active projects, weekly review: all updated automatically in the background. You see the results without opening a conversation.
Want to change anything? Just tell it.
Every 3 days, when the project is idle, it scans across metrics, customers, and channels and leaves a note with what it noticed. Always suggests. Never acts on its own.
Token-efficient by design.
Every day, git diff HEAD~1 detects exactly what changed. 1,000 files in the project, 3 changed today: it reads 3. Not a full re-read. Just the delta.
One changed file updates every dimension it touches at once: product knowledge, customer segments, metrics, skills. Automatic. Lightweight.
Every instruction becomes permanent.
Say it once: "When flagging fraud, always cross-check disposable email domains, duplicate names, and signup timing."
It writes _brain/skills/fraud.md immediately. Loaded at every future session start. Applied automatically, forever. No reminders. No editing files. No re-explaining.
Pair it with speech-to-text.
Add Wispr Flow or handy.computer and it starts feeling like...