Rizel Scarlett

Profile

Quotes

1. "The session is the story."

   Human Attention is a Scarce Resource

2. "We often throw away the one record that explains all of them: the agent session itself."

   Human Attention is a Scarce Resource

3. "When you work with coding agents, a lot of important information lives outside the final diff."

   How to Make Coding Agents Remember Past Solutions

4. "We can now produce code faster than we can understand it."

   Human Attention is a Scarce Resource

5. "Capturing our agent sessions and version controlling them ensures this rich decision history becomes a live, permanent part of the project repository."

   How to Make Coding Agents Remember Past Solutions

6. "AI is a non-sentient tool that processes historical data patterns, is prone to hallucinating, misses important context, and provides non-deterministic output."

   Your AI Agent isn't an Engineer

7. "The real value of AI developer tools includes automating boring tasks, faster prototyping, and quicker debugging, which leaves more time for creative problem-solving."

   Your AI Agent isn't an Engineer

8. "Developers juggle multiple roles - they're parents, open source maintainers, bootcamp instructors, and more."

   Your AI Agent isn't an Engineer

9. "Show what actions the Agent will take before executing them and provide detailed logs for debugging."

   Your AI Agent isn't an Engineer

10. "Technology adoption doesn't happen with one software engineer coding in solitude."

    Developer Relations is an all-company effort

11. "Developer Relations is not just a team department or an isolated function."

    Developer Relations is an all-company effort

Current Work

At Entire, Rizel is focused on the missing context around AI-assisted software work. The Entire CLI hooks into Git and agent workflows to capture prompts, transcripts, tool calls, files touched, checkpoints, and the reasoning behind commits. The project treats agent sessions as a durable software artifact, not disposable chat history.

The key idea is simple: a Git diff can tell you what changed, but it rarely tells you why an agent chose that path. Entire tries to preserve the narrative behind a change so a future developer, reviewer, or agent can ask why a line exists and recover the session that produced it.

Motherhood

Motherhood shows up in Rizel's public writing as part of her broader view of developer work: people are not abstract productivity machines. They are parents, caregivers, travelers, speakers, maintainers, and community members with real constraints outside the editor.

In Why Every Conference Needs a Nursing Room, she writes about her first solo work trip after becoming a mother and how a nursing station changed from a nice-to-have amenity into a health requirement. The piece is personal, but the argument is structural: inclusive developer events need to account for nursing parents as a basic condition of participation.

That thread connects back to her agent writing. When she argues that agents should augment developers rather than replace them, the developer she has in mind is a whole person with competing demands on time and attention. Agents are useful partly because the average developer is already multitasking across work and life.

Goose

Rizel previously worked on Goose, an open-source AI agent that runs as a desktop app, CLI, and API. Goose is built to work across model providers and connects to extensions through the Model Context Protocol. It has since moved to the Agentic AI Foundation at the Linux Foundation.

Goose is one place where her interests converge: open source, agentic workflows, MCP, and the question of how much control developers should keep as tools become more autonomous.

MCP

Her recent writing on MCP moves past the basic "tools for models" framing. In 5 Tips for Building MCP Apps That Work, she describes MCP Apps as interactive UI inside agents, with practical constraints around host context, loading states, error states, what the model can see, what the view can see, and which actions require an explicit human click.

That makes her MCP perspective unusually grounded: it is about product boundaries, safety, and user experience, not only protocol mechanics.

Agent Memory

In How to Make Coding Agents Remember Past Solutions, Rizel writes about the rare-problem tax: the low-frequency debugging problem that costs an hour every few months because nobody wrote down the path to the fix. Her answer is not always a reusable skill or a detailed runbook. Sometimes the right artifact is the captured session: what failed, what worked, what command proved it, and why it should be trusted next time.

Review Bottleneck

In Human Attention is a Scarce Resource, she argues that agentic coding shifts scarcity from code production to human review and understanding. Teams can now create more code than they can comfortably inspect. That makes provenance, summaries, session history, and checkpointed decision-making more important than ever.

Agent Framing

Her essay Your AI Agent isn't an Engineer pushes back on replacement language. The point is not that agents are synthetic coworkers. The point is that they are tools that can help developers automate boring work, prototype faster, debug faster, and preserve more attention for the human parts of engineering.

Recurring Ideas

• Agents are tools, not replacement engineers.
• Generated code makes review and understanding the hard part.
• The session is the story behind the diff.
• Useful memory is decision history tied to durable artifacts.
• Good agent integrations need hooks, readable transcripts, stable session IDs, and testable non-interactive modes.
• MCP is also about interface design, data boundaries, and human confirmation.