Dev.to
8/2/2026

Portable Agent Governance at Solo-Developer Scale: A Four-Domain Case Study
Short summary
A solo developer maintains a file-based agent governance protocol across four production projects using three files: CORE.md (fixed principles), AGENT.md (execution behavior), and SESSION_INDEX.md (cumulative memory). Rules learned from concrete failures in one project are inherited by projects in completely different domains. The approach addresses the gap between enterprise governance tools—expensive and incomplete at small scale—and the need for disciplined AI oversight.
- •Three-file protocol (CORE.md, AGENT.md, SESSION_INDEX.md) governs AI agents across four independent production projects
- •Rules from concrete failures in one domain are inherited by projects in entirely different domains
- •Enterprise governance tools are heavy and incomplete at solo-dev scale; file-based self-revising protocols fill the gap
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



