ZWISERFIT runs a real gym with 9 autonomous AI agents. 107 consecutive days. Zero humans in daily ops. Your PR becomes a production rule.
Here's how to contribute.
Three Entry Levels
π’ Level 1 β Audit: Read our public decision logs. Find a mistake our agents made that we didn't catch. Open a Discussion with "Audit:" prefix. That's it.
π‘ Level 2 β Issue: Browse issues tagged good-first-issue. Comment "Claiming." Fork, code, PR. Acceptance criteria is in every issue.
π΄ Level 3 β Independent Audit: Run our audit scripts against published logs. Cross-check agent actions vs documented constraints. Catch a deviation = most valuable contribution possible.
What Makes This Different
Every merged PR gets an entry in our public production changelog. Your commit hash documents:
- When your change activated
- Which agents it affected
- Which production rules changed
Your code runs in a live business within hours. Not in staging. Not in a sandbox. In production.
What NOT to Contribute
- Closed-source forks β Operational code is MIT. Go ahead. Forks impersonating the system are not welcome.
- Data scraping tools β We verify. We don't sell data. If your PR doesn't have a verification layer, it misses the point.
- Dashboards β MAFBE is an engine. PRs adding dashboards without audit trails get rejected.
The Repos
- 𧬠retroonto β The production constraints. 11 rules governing 9 agents. Change a rule, change the business.
- βοΈ zwiserfit-ai-store-manager β The agent framework. Tool configs, and deployment templates.
- π github.com/ZWISERFIT β All public repos.
Start Here
Fork retroonto. Read the 11 constraints. Open an Issue saying "I found gap X."
Or open a Discussion with "Question:" prefix. No contribution is too small β the agents are always watching for new signals.
The entire system is open. The audit is open. The logs are open. Your contribution becomes part of a running autonomous business.
github.com/ZWISERFIT
Top comments (0)