Sources#
Summary#
Engineering leader at Google working on Chrome and developer experience, and a long-time author/educator (web performance, JavaScript design patterns, image optimization). Through 2026 he publishes a widely-read blog series on AI-assisted software engineering that the field treats as a synthesis layer — taking the practices labs and practitioners demonstrate and giving them names and anatomies. His "Loop Engineering" essay (June 2026) is the source that named Loop Engineering and laid out its five-primitives-plus-memory structure.
The 2026 AI-coding essay series#
Osmani's blog functions as a connected vocabulary for AI-native engineering. The essays referenced from "Loop Engineering" (most not yet ingested into this wiki):
- Agent harness engineering — the environment a single agent runs inside. The wiki's home is Agent Harness Engineering.
- The factory model — the system that builds the software.
- Loop engineering — the harness "on a timer," one floor up → Loop Engineering.
- Long-running agents — why memory must live on disk, not in context.
- Intent debt — an agent starts every session cold and fills any gap in your intent with a confident guess; skills are intent written down on the outside (cf. Agentic Technical Debt).
- Comprehension debt — the gap between code-that-exists and what-you-understand, which compounds as the loop ships code you didn't write (cf. Outsource Your Thinking, Not Your Understanding).
- Cognitive surrender — taking whatever the loop returns instead of having an opinion (cf. Jagged Intelligence (Ghosts, Not Animals)).
- The orchestration tax — the human review bandwidth that caps how many agents you can actually run.
- The code agent orchestra / adversarial code review — the maker/checker split that Verification as the New Bottleneck formalizes here.
- Code review in the age of AI — "your job is to ship code you confirmed works."
Stance#
Osmani is a measured optimist on loop engineering: he calls it "a preview of how our work is going to evolve" while staying openly skeptical ("it's still early") and insistent on the cost caveat and the human's residual responsibility. His throughline across the series is that AI shifts where the leverage is — from typing code, to prompting, to designing systems that prompt — without removing the engineer's accountability for verification and understanding. "Build the loop. But build it like someone who intends to stay the engineer."
Connections#
- Loop Engineering — named the discipline and supplied its five-primitive anatomy
- Agent Harness Engineering — his "agent harness engineering" essay; loop engineering sits one floor above it
- Verification as the New Bottleneck — his code-review / adversarial-review / orchestra essays all reduce to this
- Outsource Your Thinking, Not Your Understanding — comprehension debt and cognitive surrender are facets of the understanding bottleneck
- Agentic Technical Debt — intent debt is the same compounding-drift failure mode
- Boris Cherny / Peter Steinberger — the two practitioners whose "design loops, don't prompt" framing the essay is built on
Sources#
- Loop Engineering — "Loop Engineering" (addyosmani.com, June 2026)
Cited by 13
- Agent Context Files
The cross-vendor markdown-as-control-plane pattern: repo-versioned plaintext (CLAUDE.md / AGENTS.md / SOUL.md / WORKFLO…
- Agent Harness Engineering
Patterns for scaffolding long-running LLM agents: environment design, progressive context disclosure, mechanical archit…
- Agent Loop Pattern
`/loop` (cron-scheduled) and Ralph Wiggum (backlog-draining) loops as next-generation agent primitive; AFK execution, p…
- Agentic Technical Debt
Debt that *compounds* (not just accumulates) because each agentic-coding session re-derives architectural decisions wit…
- AI Brain Fry
Kropp et al. 2026/03: mental fatigue from excessive AI oversight increases minor errors +11%, major errors +39%; cognit…
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React; CLI/desktop/web/mobile/IDE sur…
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Human-AI Accountability Redesign
HBR five-pillar prescription: span-of-control redesign, role redesign, performance management reset, decision-rights/es…
- Loop Engineering
Replacing yourself as the agent's prompter by designing the system that prompts it: a recursive-goal loop built from fi…
- Entities — People, Orgs, Tools & Projects
Map of Content for all 39 entity pages. See Home for concept domains.
- Outsource Your Thinking, Not Your Understanding
"You can outsource your thinking but not your understanding"; understanding as the non-delegable human bottleneck; know…
- Peter Steinberger
Founder of PSPDFKit turned prolific independent AI-coding experimenter (@steipete); originated the framing that loop en…
- Verification as the New Bottleneck
Fiona Fung: coding is no longer the bottleneck — verification, review, maintenance are; shift-left; TDD loses its tax;…
Related articles
- Loop Engineering
Replacing yourself as the agent's prompter by designing the system that prompts it: a recursive-goal loop built from fi…
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Vibe Coding vs. Agentic Engineering
Vibe coding raises the floor (anyone builds); agentic engineering preserves the quality bar while going faster; ">10x a…
- Agent Loop Pattern
`/loop` (cron-scheduled) and Ralph Wiggum (backlog-draining) loops as next-generation agent primitive; AFK execution, p…
- Andrej Karpathy
Co-founder OpenAI, ex-Tesla AI, Eureka Labs; coined "vibe coding," Software 1/2/3.0, "ghosts not animals," "agentic eng…
