Self-improving Compose Multiplatform UI harness. Inspection tests render every screen across a stress matrix; a deterministic Tier-1 asserter plus a vision evaluator grade them in a separate lane; recurring findings ratchet into permanent gates via a compounding feedback ledger. Ships Konsist architecture rails and a setup skill that scaffolds it all into a project.
Automated code review for pull requests using 9 specialized agents with confidence-based scoring, severity classification, and consensus validation. Automatically triggers error handling, test coverage, type design, and comment analysis based on changes. Includes optional code simplification.
Document gardening + authoring — v1.6.1 fixes the native Windows atomic-write identity false-positive (path-lstat dev:0); scan/garden/audit for Claude Code and Codex.
Lossless local byte-page memory for MCP assistants in native Gold/Helium: exact-match nonce lookup through an RGBA texture LUT; stores every conversation turn for exact recall.
Guided onboarding that connects your tools, interviews you about your business, migrates your existing projects, and builds a personalized AI workspace with context files and a self-correcting rules engine.
Triage every Claude Code session you have open and report which ones are blocked on you - the question each one asked, what only you can do, and where two sessions are duplicating work. Includes /c-assistant-voice, a spoken variant for when your hands are busy.