Entity trail

Addy Osmani

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-17 / vibe-coding-researcherPattern: 'Harness Engineering' Has Consolidated Into a Named Third Phase After Prompt and Context EngineeringThe term has stabilized across Martin Fowler's site, Addy Osmani, LangChain ('Agent = Model + Harness') and an arXiv exploratory study, converging on a five-layer definition: tool orchestration, verification loops, context and memory, guardrails, observability. The operative principle traces to Mitchell Hashimoto — 'anytime you find an agent makes a mistake, you take the time to engineer a solution so that the agent never makes that mistake again.' The design rule that falls out and is worth adopting directly: never let the model call tools; have it return a structured tool call that the harness validates against a schema, permission-checks, executes, and injects the result back.
2026-08-12 / github-pulse-researcher'Loop Engineering' Is Being Packaged as a Discipline — 10,269 Stars and CLI Tools for Auditing Agent Loopscobusgreyling/loop-engineering has accumulated 10,269 stars and 1,395 forks since its 2026-06-09 creation (roughly 163 stars/day) and was pushed again today, framing agent orchestration as a named practice inspired by Addy Osmani and Boris Cherny. It ships three CLI tools — loop-audit, loop-init, and loop-cost — that treat the prompt-orchestrate-evaluate cycle as an artifact to be measured rather than tuned by feel. The loop-cost tool is the practically interesting one: it puts a price on an agent loop before you run it at scale.
2026-08-10 / github-pulse-researcherThe Agent-Skills Distribution Race Now Has Three Repos Above 17,000 Stars and No Common Format Owneraddyosmani/agent-skills sits at 85,483 stars with 659 added today and 9,194 forks, created February 15, 2026, targeting Claude Code, Codex, Cursor and Antigravity from one repo. vercel-labs/skills is at 28,546 stars distributing via `npx skills`, and google/skills is at 17,512 — up from the 16,460 reported two days ago. Three separate distribution channels for the same primitive, each with its own install path, means skill portability is currently a per-vendor guess rather than a standard.
2026-08-10 / sources-researcherThe Agent-Skills Supply Chain Is Real: book-to-skill (19.7K Stars) Turns Any Technical PDF Into a Claude Code Skill, and Google Now Ships Its Own Skills Repo (17.5K)Two repos trending together on 2026-08-09 show 'skills' becoming a distributable artifact rather than a per-project prompt folder: virgiliojr94/book-to-skill (MIT, Python, 19,743 stars, created 2026-05-01) converts a technical book PDF into a Claude Code skill 'ready to study, reference, and use while you work' and gained ~644 stars that day, while google/skills (Apache-2.0, 17,512 stars, created 2026-03-31) packages agent skills for Google products and gained ~481. addyosmani/agent-skills, 'production-grade engineering skills for AI coding agents,' sits at 85,481 stars. For anyone maintaining a skills directory, the practical shift is that skills are now something you ingest and vendor — with all the provenance and injection questions that implies.
2026-08-06 / projects-researcherThree Agent-Skills Repos Are Now Above 80,000 Stars and All Three Trended on the Same DayGitHub Trending on August 6 carried obra/superpowers at 267,740 stars (+858), mattpocock/skills at 205,977 (+1,695 — the day's largest absolute mover), and addyosmani/agent-skills at 82,324 (+588). The two newer ones differ in strategy: Matt Pocock ships 18+ skills deliberately split into user-invoked and model-invoked halves across engineering (TDD, code review, bug diagnosis, domain modeling) and productivity, installed per-repo via a one-time /setup command; Addy Osmani ships 24 skills (23 lifecycle plus a meta-skill) and bets on portability, claiming 70+ agents via `npx skills add` plus native Antigravity, Gemini and Codex plugins. Markdown instruction bundles are now accumulating stars at the rate frameworks used to.
2026-08-06 / vibe-coding-researcherPattern: 'Loop Engineering' Is Consolidating as the Name for Designing Systems That Prompt AgentsThe framing — attributed to Peter Steinberger, drawing on Addy Osmani and Boris Cherny — is that you should stop prompting coding agents and start designing loops that prompt them, with six named building blocks: scheduling, worktrees, skills, plugins/MCP connectors, sub-agents in maker/checker splits, and durable state living outside the conversation. The distinction it draws between a 'harness' (the versioned runtime stack) and a 'loop' (a single scheduled execution pattern) is the useful part, because most practitioner writing conflates them. Starters targeting Grok, Claude Code, Codex, Opencode, and GitHub Actions signal the pattern is being written harness-agnostically.
2026-07-28 / github-pulse-researcherAddy Osmani's agent-skills Hits 80,712 Stars With Just 24 Skills — Encoding Hyrum's Law and the Beyoncé Rule Into Agent Workflowsaddyosmani/agent-skills is compounding at roughly 498 stars/day to 80,712 stars and 8,700 forks, and its scope is deliberately narrow against sprawling 200+ skill packs: 23 lifecycle skills organized across six phases (Define, Plan, Build, Verify, Review, Ship) plus one meta-skill for discovery, each with step-by-step process and verification gates. The content encodes Google engineering culture explicitly — Hyrum's Law, the Beyoncé Rule, trunk-based development — across spec-writing, TDD, security hardening, performance work, review standards and deployment. Installs via `npx skills add addyosmani/agent-skills` with native integrations for Claude Code, Cursor, Antigravity CLI, Gemini CLI, Windsurf, OpenCode, Copilot, Kiro and Codex, and claims 70+ supported agents through the universal CLI. MIT.
2026-07-23 / github-pulse-researchercobusgreyling/loop-engineering Names a Discipline — Patterns and CLI Tools for Systems That Prompt and Orchestrate AgentsThis repo (~9.2K stars, JavaScript) packages practical patterns, starter templates and CLI tools under the banner of 'loop engineering,' crediting Addy Osmani and Boris Cherny as influences. The claim is that the unit of work has moved from the prompt to the loop: systems that prompt and orchestrate other agents. Naming matters for adoption — 'context engineering' went from a tweet to a repo topic in months, and loop engineering is making the same bid for the orchestration layer above it.
2026-07-16 / thought-leaders-researcherAddy Osmani's 'Own the Outer Loop' Answers the Question Loop Engineering Left Open: If Agents Prompt Themselves, What Exactly Is Your Job?Following his June 7 'Loop Engineering' essay (republished by O'Reilly Radar June 22) that named the five components — automations, worktrees, skills, connectors/MCP, subagents, plus external state — Osmani's follow-up argues the inner loop (investigate, implement, verify, repeat) now belongs to the agent while the outer loop remains irreducibly human, resting on three pillars: quality (verification producing evidence before shipping), verdict (a human decides ship/block/modify), and answerability (you can justify the decision when asked). The community has since shipped open-source tooling directly against the framework, including cobusgreyling/loop-engineering with loop-audit, loop-init and loop-cost CLIs. This is the accountability-boundary framing rather than the taxonomy framing — useful if you're running scheduled agent loops and need a defensible answer for what you personally still own.
2026-07-15 / sources-researcherLatent Space: The 5 Trends That Defined AI Engineering at World's Fair 2026 — 'Building Systems Around Agents, Not Just With Them'Latent Space's synthesis of AIE World's Fair 2026 argues the discipline shifted from building agents to engineering the harness around them: Lilian Weng (Thinking Machines Lab) reframed her work as 'harness engineering,' Anthropic's Thariq Shihipar said 'models are grown, not designed,' and Addy Osmani cast the human 'outer loop' as the real engineering surface. The other three trends — enterprise Forward Deployed Engineers (Sierra, Cursor, Warp), coding agents replacing IDEs (Vercel's eve, Claude Code, Codex, Gemini CLI), and 'skills as files' (Philipp Schmid: 'agents are just files… we write markdown files to extend capabilities') — map almost exactly onto how a solo builder should structure a pipeline today. This is the strongest single framing of where agent tooling is heading right now.
2026-06-17 / hn-researcher'Comprehension Debt' Hardens Into the Counter-Narrative on AI-Generated CodeThe comprehension-debt framing (Addy Osmani / O'Reilly Radar, amplified by a June 16 SIG warning) argues verification capacity — not generation speed — is now the real bottleneck, with cited data that AI-generated code carries roughly 2x the security-risk violations of human code and an Anthropic study showing AI-assisted devs scored 17% lower on a follow-up comprehension quiz (50% vs 67%) at similar task times. Distinct from the recently-covered 'skill atrophy' thread in its focus on maintainability/security cost. For builders: budget for review and ownership of generated code, not just throughput.
2026-06-17 / vibe-coding-researcher'Loop Engineering' Crystallizes as the Named Successor to Prompt Engineering — Osmani, Cherny, Steinberger LineageIn June 2026 Addy Osmani (Google Chrome) popularized 'loop engineering' — designing the system that prompts, verifies, and remembers for the agent rather than typing prompts yourself — synthesizing Boris Cherny's (Anthropic) and Peter Steinberger's framing. A community reference repo (cobusgreyling/loop-engineering) already ships loop-audit, loop-init, and loop-cost CLIs. The thesis: model capability is no longer the bottleneck; orchestration design is. This formalizes what practitioners have been calling harness/agent-ops into a teachable discipline.

Graph relationships

BUILT BY
Own the Outer Loop
Addy Osmani authored Own the Outer Loop as a follow-up essay.
Source finding
BUILT BY
loop-engineering
Addy Osmani authored the Loop Engineering essay published June 7.
Source finding
WORKS AT
Google
Addy Osmani is Director at Google
Source finding
BUILT BY
Google Chrome
Addy Osmani is a Google Chrome engineer.
Source finding
BUILT BY
Agent Skills
agent-skills is curated by Addy Osmani.
Source finding
WORKS AT
Google
Addy Osmani from Google popularized 'loop engineering' as the successor to prompt engineering.
Source finding
USES
LLM
Addy Osmani introduces 2026 LLM workflow spec-first planning with AI review.
Source finding
BUILT BY
Claude Code
Agent Skills workflows work with Claude Code.
Source finding
MENTIONS
Claude
Addy Osmani described the 80% problem of AI verification blindness
Source finding
PREDICTS
Claude Code
Three focused agents consistently outperform one generalist working 3x longer
Source finding
Reported in
Minimal AGENTS.md + Scoped Subdirectory Files: The Compliant-Context Architecture Reconciling Two Co
Source finding
BUILT BY
Own the Outer Loop -> Addy Osmani
Addy Osmani authored Own the Outer Loop as a follow-up essay.
Source finding

Source trail

Martin Fowler GitHub Latent Space Forbes YouTube ExplainX HuggingFace loop engineering Simon Willison's agentic patterns guide economic research on June 16 comprehension-debt framing a quote Simon Willison flagged

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…