Solomonar is an open-source framework for developers who want AI coding agents to work from agreed specifications, follow focused tasks, and pass independent reviews.
When requirements live in chat history, agents can lose context and reviewers have no shared definition of done. Solomonar keeps agreed behavior in version-controlled specifications, gives each task acceptance criteria, and separates implementation from review.
Spec-driven development means agreeing on what software should do before asking an agent to implement it, then checking the result against that specification. The framework connects specifications, agent skills, and orchestration and review in one workflow.
The design targets Claude Code and Codex, with a language-agnostic core. Runtime adapter installation and project initialization are still in development. See the current CLI commands and workflow examples with agent prompts.
In Romanian folklore, a solomonar learns from a single book and rides a dragon to command the storms. Your project has its book (specs), spells (skills), and dragons (agents).
