Stanford course teaches AI-native engineering with open-source maintainer reviews and coding agents
CS146S students learn to navigate unfamiliar repos, propose changes, and defend patches to maintainers, using coding agents as part of the curriculum.
1 source · cross-referenced
- Stanford’s CS146S course now includes open-source maintainers reviewing student code as part of an AI-native engineering curriculum.
- Students must learn an unfamiliar repository, propose a change, and explain the patch to maintainers who may merge it.
- The course syllabus and assignments are designed around real-world agent workflows, including context engineering and code review.
- Monaco’s engineering team contributed tooling decisions and agent workspace setups used in the course.
- The interview highlights how coding agents are integrated into teaching and industry practices.
In the second iteration of Stanford’s CS146S course, instructor Mihail Eric incorporated open-source maintainers into the grading and feedback process, requiring students to learn an unfamiliar repository, propose a change, and defend their patch to maintainers who could merge it. The course is part of a broader push to teach AI-native engineering, where coding agents are not just tools but integral to the workflow.
The ten-week syllabus and assignments are designed around agent-assisted development, including lessons on context engineering for large codebases and the skills needed to understand and review code diffs. Eric, who teaches "The Modern Software Developer" at Stanford and leads AI at Monaco, shared how the course’s structure evolved after consulting with his engineering team and observing real-world agent deployments.
Monaco’s engineering team contributed practical insights into tooling decisions, including the setup of agent workspaces and the integration of code review tools. The company also provided access to its internal processes, such as giving an on-call agent access to Datadog and GitHub, which informed the course’s emphasis on security, dependencies, and permissions.
The conversation highlighted the role of background agents in modern software development, drawing parallels to Stripe’s reported use of background coding agents. It also touched on the challenges of scaling agentic workflows, such as when code review becomes a bottleneck, and the importance of making repositories "agent-ready" to close the loop between generation and deployment.
Eric emphasized a pragmatic approach to tooling: "You buy to learn, and you build to keep," reflecting a balance between adopting existing tools and building custom solutions to maintain competitive advantage. The course and Monaco’s practices illustrate how organizations are moving from personal adoption of AI tools to enabling team-level autonomy in agentic environments.
- Nov 11, 2026 · Latent Space — swyx
OpenAI demonstrates voice steering, cloud handoffs, and personal agents for Codex at AI Engineer Code Summit
Trust79 - Nov 11, 2026 · The Verge — AI
OpenAI to launch verified AI Business Agents in ChatGPT for US e-commerce and service businesses
Trust79 - Nov 10, 2026 · Latent Space — swyx
NaiveAI releases Naive-N0.5-Flash coding model with ultrafast token throughput
Trust74