humanagentworld
Execution
Human-friendly agent toolkits
We don't rebuild the agent. We help you see, steer, and review what it does — and turn every run into feedback a human can actually use.
- TokenFlow
- Runlight
- Agent-in-Chat
A Renaissance for the age of AI
The Renaissance of the human mind
The first Renaissance turned our gaze from gods to humans. The age of AI must turn it from agents back to humans. We don't just make AI stronger — we make you stronger.
- 0
- core directions
- local-first
- privacy stance
- human-led
- design center
Everyone is racing to make the agent smarter. We think the more interesting question is how to make the human smarter.
Deep research can gather and summarize. But genuine understanding — building judgment, forming intuition, learning to think — is still slow. RenaissanceMind is a family of tools and research aimed at one thing: human evolution in the age of AI.
不只是让 AI 更强，而是让人更强。
01 — The System
Three directions, one center: you.
Information flows through agents — but it always begins and ends with a person.
human
Context
The model of you
What you know, how you learn, how you judge, how you like things explained — and which data must stay local. The substrate that lets AI understand a person over time.
- Cognition Modeling
- Preference
- PrivacyContext
worldagenthuman
Cognitive Augmentation
Knowledge → human-preferred form
Reshape the world's information into representations a mind can absorb, question, and build on — not just another summary to skim.
- DrawAI
- MindCanvas
- Knowledge Presentation
02 — Projects
Small, sharp tools — with the human in the loop.
Small, sharp tools — not another monolith. Each one returns control and clarity to you.
TokenFlow
Your AI agents finally have a meter.
Private, local-first token accounting for the agents you actually use. Scans locally, aggregates locally, uploads only usage metadata — prompts, responses, source file paths, and session IDs stay off the server.
View on GitHub →Runlight
Passive observability for AI coding sessions.
A real-time status light for your agents — running, waiting, stuck, done, or failed. See it at a glance, across every project.
runlight.renaissancemind.ai →Agent-in-Chat
Put the agent inside your conversation.
Discuss as usual; @-mention the agent when you need it. It acts with the recent chat as context — summaries, lookups, alignment, light tasks.
agent-in-chat.renaissancemind.ai →DrawAI
Make any knowledge image editable.
An image-to-editable-SVG/PPTX reconstruction pipeline. Inspect, edit, and reuse AI-generated or existing visuals instead of treating them as flat pixels.
Knowledge Presentation →ME / Context
A personal context layer that remembers.
Not a memory dump — a structured model of your knowledge, preferences, and privacy boundaries, in service of learning, thinking, and review.
In researchMindCanvas
Brainstorm, note, and think with AI.
A surface where thinking becomes operable context — where notes and conversations turn into structure you can act on.
Concept
The point of all of it
From a single curious mind
to a connected cosmos of understanding.
Tools come and go. What endures is a person who can see further, think more clearly, and judge more wisely — with intelligence as an instrument, not a replacement.
03 — How we build
Four principles we don't bend.
-
01
Human-first, always
Every feature is judged by one question: does it make the person more capable? If it only makes the agent louder, it doesn't ship.
-
02
Local-first & private
Your prompts, files, and behavior are yours. We process on your machine first, and you decide what — if anything — ever leaves it.
-
03
Observe, don't replace
We make the agent visible and reviewable. The human keeps the wheel; the tool keeps the receipts.
-
04
Real over mock
Real repos, real commands, real output. We'd rather show something honest and small than something polished and fake.
Join the Renaissance of the mind.
Follow the work, open an issue, or just watch the projects grow. The gaze is returning to the human — come help point it there.