Autonomous orchestration framework for Claude Code with MemPalace-inspired memory (4-layer stack, 818-token wake-up), parallel-first Agent Teams (6 teammates), Aristotle First Principles methodology, and 4-stage quality gates. 925+ tests, 22 active hooks, automatic learning pipeline.

Eval framework. Define correct, test against it, get results.

Ship evals before you ship features.

AI-augmented QA platform for spec-driven development and testing, RAG-grounded analysis, eval-driven development and contract validation across Python, Go, Rust and Solidity.

A hands-on learning repository exploring Spec-Driven Development (SDD) for building deterministic AI systems. Covers specs, evaluation loops, patterns, experiments, and failures to bridge theory with real-world AI engineering practices.

Autonomous skill improvement loop for Claude Code plugins — inspired by Karpathy's autoresearch. Modify → evaluate → keep/discard → repeat until convergence. Zero-touch quality iteration at scale.

Most AI plugins hope they work. These prove it. Eval-driven Claude plugins for product teams.

🗂 Coordinate multiple AI agents using shared plans and structured tasks for efficient team-based coding with Claude Code Agent Teams.

Define, measure, and enforce code correctness with Eval-Driven Development, ensuring every probabilistic system ships with automated proof of quality.