Prompt Engineering & Management

Reflective prompt evolution optimizer using natural language reflection and Pareto frontier learning. Outperforms reinforcement learning for prompt optimization. Integrated with DSPY and MLflow. MIT licensed.

Local-first MCP server for explicit-budget context selection, content-addressed exact recovery, and auditable Context Receipts. Apache 2.0 licensed.

Staged prompt workflow and CLI that turns a product idea into a PRD, technical design, and `AGENTS.md` instruction files for AI coding agents. MIT licensed.

Self-learning browser infrastructure for AI agents that compiles a site's navigation into deterministic per-site CLI commands (Apache-2.0, TypeScript).