Open-source studio
Five projects. One studio.
A human-in-the-loop measurement family and a pair of developer tools — each ships independently.
Human-in-the-loop AI, measured properly.
Sixteen React primitives for human-in-the-loop agentic UIs. Each primitive is the physical embodiment of a specific claim from the paper.
npx shadcn@latest add https://www.hitlkit.dev/r/hitl-card.jsonA measurement instrument for multi-step research agents.
Humans score, not LLMs. LLM-as-judge is only an opt-in pre-fill, flagged on every score.
npx @eval-kit/core init my-evalsStructured tagging primitives for human-in-the-loop annotation workflows.
Most “tagging” features in HITL tools are unstructured strings — you type a label, it gets stored next to the decision, and then nobody can aggregate or score across them later.
pnpm add @tag-kit/core @tag-kit/uiA Claude Code skill-building framework.
Three pluggable ingestors (MDX lessons, Jupyter .ipynb / MyST .md, and a one-file extension pattern for any source format) feed a typed pipeline that compiles each pattern into a SKILL.md and atomically writes it to ~/.claude/skills/.
git clone https://github.com/akaieuan/collapse && pnpm install && pnpm devLive observability, guided skills, and an agent (MCP) surface for Blender → glTF pipelines.
Hologram watches a glTF asset pipeline and streams what's happening to a local dashboard in real time — including the tool calls your AI coding agent is making right now.
uvx --from hologram-gltf hologram dashboard