Mneme HQ
Enforces approved architectural decisions before code is generated, providing guide, warn, or block verdicts traced to source.
Mneme HQ is an engineering governance layer for AI coding agents that enforces approved architectural decisions before code is generated. It solves the problem of architectural drift caused by AI-generated code outpacing human review. The product targets platform engineering and DevEx teams in mid-market to enterprise organizations. It is delivered as an open-source CLI and local package, with a structured decision corpus stored in the repository, and integrates with CI and agent hooks.
Key features
- Turns ADRs into enforceable verdicts
- Applies decisions before code generation
- Provides guide, warn, or block verdicts
- Traces each verdict to source
- Stores decisions as repo-native JSON
- Integrates with Claude Code hooks
- Supports CI enforcement via GitHub Actions
- Scopes rules per team or repository
- Resolves conflicts with precedence engine