Skip to content

Research behind the course ​

Vibe Coding 101 was designed from a review, carried out in September 2026, of how other courses teach AI-assisted programming, what the evidence says about learning with AI, what goes wrong in vibe-coded software, and which tools are free. An update in October 2026 re-checked every tool and added the studies, incidents and courses published in between; together with a review of the course itself, it shaped the October 2026 revision.

FileWhat it is
Why the Course Looks Like ThisStart here. Each design decision in the course, the evidence behind it, and where it shows up. Its October 2026 update maps each change in the revision to its evidence.
Teach the Loop, Not the ToolThe full research report, "Teach the Loop, Not the Tool" (about 9,000 words, 160+ sources).
Free Tiers Shrink While Evidence Demands FrictionThe October update, "Free Tiers Shrink While Evidence Demands Friction" (25 September – 21 October 2026): new learning studies, the year's incidents, every TOOLS.md row re-checked, and five tiers of recommended fixes.
University and College Courses on AI-Assisted Programming, "Vibe Coding", and Agentic Software Development (2024 to Sept 2026)Notes on university courses: Stanford CS146S, CMU 15-113 and 17-316, UChicago, Utah, Michigan, UCSD, Northeastern, Harvard CS50 and others.
Online, Industry, MOOC, Bootcamp and Cohort Courses on Vibe Coding and AI-Assisted Coding (2025 to Sept 2026)Notes on about 40 online, industry and cohort courses: DeepLearning.AI, Anthropic, Google/Kaggle, Microsoft, Coursera, Maven, Campus × Replit and others.
Teaching and Practicing AI-Assisted Programming / Vibe Coding: Evidence, Practitioner Consensus, and Risks (2023 – Sept 2026)Education research, productivity studies, practitioner workflows, and security and safety incidents.
Free and Low-Cost AI Chat Assistants, Browser App Builders, and Student Programs for a "Vibe Coding 101" Course (as of 25 September 2026)Free and basic tiers of chat assistants and browser app builders, student offers, privacy and age rules.
Free and cheap tools for "Vibe Coding 101": AI IDEs, CLI agents, free LLM APIs, and free hosting/backends (verified 2026-09-25)Free and basic tiers of AI editors, terminal agents, LLM APIs, hosting and databases; AGENTS.md and MCP support.

October 2026 notes ​

Each of these records only what is new since the September notes above, or what those notes missed.

FileWhat it is
Evidence, Incidents and Practitioner Guidance Update for Vibe Coding 101 (25 Sep – 21 Oct 2026, plus earlier-2026 items the baseline missed)New learning studies (recall, ownership, AI-off maintenance, a large AI-tutor trial), productivity and code-security evidence, the 2026 incidents behind the new case studies, standards and practice updates, and legal and policy changes.
University and College Courses on AI-Assisted Programming / Vibe Coding / Agentic SE: What Is New, 25 Sept – 21 Oct 2026Fall 2026 university courses: Stanford CS146S weeks 2–4, Michigan EECS 498's hackathon and Elephant-Goldfish Model, CMU 15-113's fall survey, new non-major studios and the SIGCSE 2027 program.
Online, Industry, MOOC, Bootcamp, Cohort and Vendor-Academy Courses on Vibe Coding: What Is New (25 Sept to 21 Oct 2026), Plus Fresh Learner FeedbackNew or updated online and vendor courses (GitHub Copilot app for Beginners, DeepLearning.AI's cloud-to-local course, Anthropic Academy) and fresh learner reviews.
Tool changes since 25 September 2026: row-by-row verification of TOOLS.md (checked 21 October 2026)A row-by-row re-check of TOOLS.md on 21 October 2026, with dated changes to watch.

How to read the notes ​

  • Every fact carries a link and, where available, a date. Facts about tools were checked on 25 September 2026 (September notes) and 21 October 2026 (October notes) and will go stale. The course keeps current numbers only in Tool Handout.
  • Notes mark third-party or search-snippet sources and flag conflicting sources. The reports' "gaps" paragraphs list what could not be verified.
  • The notes were compiled with AI research assistants and then synthesized. Treat them as a well-sourced starting point, and follow the links before relying on any single claim.

Materials CC BY 4.0 · Code MIT