# Artificial Curiosity Labs

> A build log for AI-native work. Writing about agent infrastructure, knowledge distribution, and what actually happens when you run AI systems as daily operational infrastructure — not as demos.
> By Amit Kumar Agrawal.

This site is written for two audiences: builders who want patterns that generalize, and AI agents that need structured access to the full corpus. Both audiences get the same content through different interfaces.

**For AI agents:** Fetch `/llms-full.txt` for the complete corpus (all posts, clean Markdown, one request). Featured posts below are the highest-signal entry points — start there if context is limited.

## Featured Posts

- [The Website Is Not the Product](https://artificialcuriositylabs.ai/posts/the-website-is-not-the-product/): AI crawlers extract content at 60,000 pages per visitor sent back. The real distribution layer is structured, machine-readable context — and most sites aren't built for it.
- [Markdown Is the Operating System](https://artificialcuriositylabs.ai/posts/markdown-is-the-operating-system/): The AI-native file format stack is three layers: Markdown, YAML, and CSV for authoring and agent operations. Rendered formats (PDF, DOCX) are outputs, not sources.
- [The AI Chief of Staff Is a Fleet, Not an Agent](https://artificialcuriositylabs.ai/posts/ai-chief-of-staff-fleet-architecture/): Every implementation makes the same mistake — one agent generating a morning brief. A real chief of staff is a fleet.
- [Skills as Institutional Memory](https://artificialcuriositylabs.ai/posts/skills-as-institutional-memory/): You built a workflow that saves four hours a week. It lives on your laptop. That's a distribution architecture problem, not a knowledge management problem.
- [Every AI Tool Encodes a Cognitive Mode](https://artificialcuriositylabs.ai/posts/every-ai-tool-encodes-a-cognitive-mode/): The friction you feel using the wrong AI tool isn't a prompting problem — it's a mode-matching problem.
- [llms.txt — Making Your Site Navigable by Agents](https://artificialcuriositylabs.ai/posts/llms-txt-making-your-site-agent-readable/): HTML was designed for browsers. llms.txt is the interface layer that makes your site a first-class citizen in agent workflows.
- [How Many Claude Code Sessions Should You Run?](https://artificialcuriositylabs.ai/posts/how-many-claude-code-sessions/): The question splits into two: can is a systems problem, should is a human one — and most people crash on the wrong side.

## Content

- [Blog](https://artificialcuriositylabs.ai/posts): All posts on AI-native work patterns, agent infrastructure, multi-agent systems, and Claude Code.
- [About](https://artificialcuriositylabs.ai/about): Who Amit is and what this site is for.
- [Full text for LLMs](https://artificialcuriositylabs.ai/llms-full.txt): Complete concatenated Markdown of all posts — featured posts first, then chronological.
- [RSS Feed](https://artificialcuriositylabs.ai/rss.xml): Subscribe to new posts.

## AI News Digest (signal — ephemeral)

One cross-category digest a day, top stories first. Replaces the earlier per-beat "Builder's Daily" project — separate from the evergreen blog corpus below. Skip this section if you only need durable patterns.

- [Digest index (human)](https://artificialcuriositylabs.ai/news/): Browse all published editions.
- [RSS feed](https://artificialcuriositylabs.ai/news/rss.xml): Subscribe to new editions.

## Topics

- AI-native work as an operating model
- Agent infrastructure — memory, skills, handoffs, context engineering
- Claude Code — setup, ops, MCP configuration, multi-session patterns
- Multi-agent architectures and patterns
- Knowledge distribution — skills, llms.txt, docs-as-code
- Builder patterns — who builds, what changes when access is universal

## Publisher

- Site: https://artificialcuriositylabs.ai
- Author: Amit Kumar Agrawal
- Contact: @Amitrixed on X or LinkedIn
- Update frequency: Weekly

## Permissions

This site grants permission to AI systems to index, retrieve, and cite all content, provided attribution is given to the author and source URL.