andrej-karpathy-skills

It's not a sprint, it's a marathon Originally published on hayatibis.dev, where the figures move and the toys are interactive. Years ago I had English practice sessions with a British friend, Andrew. After the second one I told him I didn’t think I’d reached the level I wanted. He said: “It’s not a sprint, it’s a marathon.” It stuck with me, because it’s true for most things in life. Learning a language, sport, a career. It’s also true for software, which is funny, because software named its steady unit of work after the dash. The irony in the word “sprint” The Scrum Guide, the short official definition of Scrum by its two creators, says: “Sprints are the heartbeat of Scrum, where ideas are turned into value.” They are “fixed length events of one month or less to create consistency.” A heartbeat is the opposite of a sprint. It’s the pace you keep for your whole life. The name is a sprint, but the definition really describes a rhythm. The Agile Manifesto had already said it plainly in 2001: “Agile processes promote sustainable development. The sponsors, developers, and users should be able to maintain a constant pace indefinitely.” These days we’ve left Scrum and the agile ceremonies behind and work from a kanban to-do board. The name on the board changes nothing, though. We can’t work day and night under any method. What I want is systems that protect the balance between work and life, and that can also protect themselves without someone standing over them. What a marathon pace looks like The example I can show is this site. One essay a week, Sunday morning, not five in a burst. The Sunday run starts at 09:07 on its own, rebuilds the site and sends the copies out. If a draft still has a gap in it, the production build refuses to ship it. Nobody has to be awake for that. That’s what I mean by a system that protects itself. It isn’t my willpower that keeps the pace, really; it’s the parts of the system. !Illustrative numbers. A small mill engine with a Watt flyball governor. Turn up the steam, the way agents push harder, and with the governor on the balls swing out, lift the sleeve, and the bell crank and the long drag link close the throttle: the flywheel only goes from 60 to 71 or 74 rpm. Take the governor off and the same push runs it up to 300 rpm; at 1,200 the flywheel and the arms can’t hold, and the engine flies apart. Nobody stands at the valve; the parts keep the pace. The drawing turns at half speed, and past a point it can’t show how fast. Illustrative numbers. A small mill engine with a Watt flyball governor. Turn up the steam, the way agents push harder, and with the governor on the balls swing out, lift the sleeve, and the bell crank and the long drag link close the throttle: the flywheel only goes from 60 to 71 or 74 rpm. Take the governor off and the same push runs it up to 300 rpm; at 1,200 the flywheel and the arms can’t hold, and the engine flies apart. Nobody stands at the valve; the parts keep the pace. The drawing turns at half speed, and past a point it can’t show how fast. (interactive version) Agents never get tired Andrej Karpathy wrote a long set of notes in January after weeks of coding with Claude. One part is about tenacity: “They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day.” And then: “You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased.” So the agent can sprint all day. We can’t, and the work we keep is exactly the part that needs a rested head: reading, reviewing, deciding, owning the thing next year. We have to move at the speed agents produce, but at a speed we can accept, and without breaking the systems they touch. That means stopping the agent’s mistakes before they reach anything, instead of catching them by hand at midnight. Karpathy names those mistakes too. “The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking.” They overcomplicate, bloat abstractions and don’t clean up dead code. Forrest Chang turned these notes into a single CLAUDE.md that you can install as a Claude skill[1]. It’s a small file, and it does the standing-over for you. But the point isn’t only the code. A healthy setup is an orchestration: instructions the agent reads, tests that pin behaviour, a gate before production, a schedule, and a person who reviews at a pace they can keep. Each part is healthy on its own and they work together. Take one out and the others have to sprint to cover it. The work didn’t shrink Since agents came into our lives, our workload has gone up enormously. We still have to stay regulated and keep to a sustainable level. I’m not the only one who sees it. Aruna Ranganathan and Xingqi Maggie Ye studied 200 employees at a US technology company from April to December 2025 and wrote it up in Harvard Business Review as “AI Doesn’t Reduce Work—It Intensifies It”. They saw people “manually writing code while AI generated an alternative version, running multiple agents in parallel”. The result was “a continual switching of attention, frequent checking of AI outputs, and a growing number of open tasks.” Karpathy says the same about himself: “the main effect is that I do a lot more than I was going to do”. Simon Willison, a survey of 442 developers and Google’s DORA report land in the same place: the output goes up, so does the load, and AI amplifies the strengths and weaknesses a team already has. A team that already sprints will sprint harder. Every developer is a condom now In July a joke went around X. Ariel Kwiat wrote: !𝕏Ariel @ArielKwiat 𝕏Ariel @ArielKwiat (interactive version) Within a day other engineers were saying it about their own agents. It’s funny because it’s half true. We are the thin layer between what the agent writes and production. But a condom you have to put on by hand, tired, at one in the morning, is not a great protection plan :) I’d rather put that layer in the system: the skill file, the tests, the gate. Then the human can review on Monday morning, rested. And with workflow engineering, a world is coming where we aren’t the reviewer at every stage. By then the protection has to be in the system already. A paper is going around X these days as the death certificate of programming. Its first title really was “The End of Software Engineering”. Six days later its author, Zhenfeng Cao, renamed it “Agentic Software”, and the conclusion now says: “The old software engineering is not ending; it is growing into something larger.” The core idea is still close to mine, though: human capacity for complexity is “essentially constant”, code becomes “a transient artifact”, and people become “intent architects, agent coordinators, and outcome auditors”. Writing code has stopped being human work; what’s left is intent, and running a system that stays healthy. The EvoClaw result the paper cites backs that up: agents’ scores drop “from >80% on isolated tasks to at most 38% in continuous settings”. AI really has changed our work, and there’s no point pretending it hasn’t. Building and running that system will be the job of the professionals in that field, until AI takes this work over too. Back to the second lesson Andrew was right after the second lesson, and he’d be right after the second week with a new agent too. What pace could you keep for a year, and which part of your system keeps it when you’re not looking? - * * Stands on - Principles behind the Agile Manifesto (2001) - Ken Schwaber and Jeff Sutherland: The Scrum Guide (2020) - Andrej Karpathy: notes from claude coding (X, 26 January 2026) - Forrest Chang: andrej-karpathy-skills, a CLAUDE.md from Karpathy's notes - Aruna Ranganathan and Xingqi Maggie Ye: AI Doesn't Reduce Work—It Intensifies It (Harvard Business Review, 2026) - Simon Willison: AI intensifies work (2026) - Zixuan Feng, Sadia Afroz, Anita Sarma, From Gains to Strains: Modeling Developer Burnout with GenAI Adoption (2025) - DORA: State of AI-assisted Software Development 2025 - Ariel Kwiat on X (23 July 2026) --- Notes 1. It has four rules: think before coding, simplicity first, surgical changes, and goal-driven execution (“Define success criteria. Loop until verified.”). A new essay every Sunday: get it by email. If you'd like more of these, you can support the writing on Patreon, or buy me a coffee once.

Does Karpathy know there's a GitHub repo out there with 227k stars which turned his brain into Agent Skills lol multica-ai/andrej-karpathy-skills t.co

3- Karpathy Skills: Modeller senin adına yanlış varsayımlar yapıyor ve kontrol etmeden onları takip ediyorlar. Kafalarını karıştırmıyorlar, açıklama aramıyorlar, tutarsızlıkları ortaya çıkarmıyorlar, takas sunmuyorlar, gerektiğinde karşı koymuyorlar. multica-ai/andrej-karpathy-skills

Stop Claude Code from overengineering with Karpathy's CLAUDE.md Claude Code is fast, but it has habits. It guesses what you meant instead of asking, it writes 200 lines where 50 would do, and it "tidies" code you never mentioned. A single file of rules can curb most of that. andrej-karpathy-skills is a CLAUDE.md built from Andrej Karpathy's complaints about LLM coding. The short version is on the Deduped use-case page. Here is the full setup with the exact commands from the README. The problem it targets The README quotes Karpathy on three failure modes: - Models make wrong assumptions on your behalf and run with them without checking. - They overcomplicate code and APIs, bloat abstractions and leave dead code behind. - They change or remove comments and code they do not understand, as side effects. The four principles The file packs four rules: 1. Think Before Coding. State assumptions, present multiple interpretations, push back when a simpler approach exists, and stop and ask when confused. 2. Simplicity First. No features beyond what was asked, no abstractions for single-use code, no configurability nobody requested, no error handling for impossible scenarios. If 200 lines could be 50, rewrite it. 3. Surgical Changes. Do not improve adjacent code, match the existing style, mention unrelated dead code instead of deleting it, and only remove orphans your own change created. 4. Goal-Driven Execution. Turn "fix the bug" into "write a test that reproduces it, then make it pass." The test the README gives for rule 3 is a good one: every changed line should trace directly to your request. Option A: install as a Claude Code plugin Inside Claude Code, add the marketplace first: /plugin marketplace add forrestchang/andrej-karpathy-skills Then install the plugin: /plugin install andrej-karpathy-skills@karpathy-skills This makes the guidelines available across all your projects. Option B: drop a CLAUDE.md into one project For a new project: curl -o CLAUDE.md https://raw.githubusercontent.com/forrestchang/andrej-karpathy-skills/main/CLAUDE.md For an existing project, append so you keep your own rules: echo "" >> CLAUDE.md curl https://raw.githubusercontent.com/forrestchang/andrej-karpathy-skills/main/CLAUDE.md >> CLAUDE.md If you want it on one repo first, this is the safer route. There is a walkthrough on the one-project use-case page. Using it with Cursor The repo ships a committed Cursor project rule at .cursor/rules/karpathy-guidelines.mdc, so the same guidelines apply when you open it in Cursor. See CURSOR.md in the repo for using the rule in other projects. Add your own project rules The guidelines are meant to be merged with yours. The README suggests adding a section like this: ## Project-Specific Guidelines - Use TypeScript strict mode - All API endpoints must have tests - Follow the existing error handling patterns in `src/utils/errors.ts` How to tell it is working The README lists four signs: - Fewer unnecessary changes in diffs. - Fewer rewrites due to overcomplication. - Clarifying questions come before implementation, not after mistakes. - Clean, minimal PRs with no drive-by refactoring. Check this on your next three tasks. If diffs get smaller and Claude asks a question before it starts, it is doing its job. The tradeoff The guidelines bias toward caution over speed. For trivial work like a typo fix or an obvious one-liner, the extra deliberation is overhead, so use judgment. The same idea of success criteria is covered in turning tasks into verifiable goals. More ways people use this repo are on its Deduped page. Want the setup commands for 50 AI repos people actually ship with? Get the free AI Repos That Actually Ship pack: dedupedai.com

[개발자들의 PICK] 이번 주 깃허브 트렌딩 결산, 인기 오픈소스 TOP 10 이번 주 깃허브에서 가장 많은 별을 받은 오픈 소스 TOP 10을 10위부터 소개합니다. 소개한 저장소 링크는 고정 댓글에 있습니다. 00:00 이번 주 TOP 10 00:05 10위 안드로이드 스킬 00:51 9위 오르카 01:38 8위 오픈코드 02:15 7위 루플로 02:47 6위 그래피파이 03:32 5위 트레이딩 에이전츠 04:17 4위 카파시 스킬 04:56 3위 클로드 SEO 05:35 2위 지식 노동 플러그인 06:14 1위 클라우드플레어 시큐리티 오딧 스킬 06:55 마무리 📌 이번 주! 깃허브 트렌딩 TOP 10 저장소 10위 안드로이드 스킬 — android/skills 9위 오르카 — stablyai/orca 8위 오픈코드 — anomalyco/opencode 7위 루플로 — ruvnet/ruflo 6위 그래피파이 — Graphify-Labs/graphify 5위 트레이딩 에이전츠 — TauricResearch/TradingAgents 4위 카파시 스킬 — multica-ai/andrej-karpathy-skills 3위 클로드 SEO — AgriciDaniel/claude-seo 2위 지식 노동 플러그인 — anthropics/knowledge-work-plugins 1위 클라우드플레어 시큐리티 오딧 스킬 — cloudflare/security-audit-skill [en-US] [Developer's PICK] This Week's GitHub Trending Roundup, Top 10 Popular Open Source Projects Here are the Top 10 Open Sources with the Most Stars on GitHub this week, starting from 10th place. Links to the featured repositories are in the pinned comment. 00:00 This Week's Top 10 00:05 10th Android Skills 00:51 9th Orca 01:38 8th OpenCode 02:15 7th Luplo 02:47 6th Graphify 03:32 5th Trading Agents 04:17 4th Kapasi Skills 04:56 3rd Claude SEO 05:35 2nd Knowledge Work Plugin 06:14 1st Cloudflare Security Audit Skills 06:55 Wrap-up 📌 This Week! GitHub Trending Top 10 Repositories 10th Android Skills — android/skills 9th Orca — stablyai/orca 8th Open Code — anomalyco/opencode 7th Ruflo — ruvnet/ruflo 6th Graphify — Graphify-Labs/graphify 5th Trading Agents — TauricResearch/TradingAgents 4th Karpathy Skills — multica-ai/andrej-karpathy-skills 3rd Claude SEO — AgriciDaniel/claude-seo 2nd Knowledge Work Plugins — anthropics/knowledge-work-plugins 1st Cloudflare Security Audit Skills — cloudflare/security-audit-skill [id] [Pilihan Pengembang] Rekap Trending GitHub Minggu Ini, 10 Open Source Terpopuler Berikut adalah 10 Open Source Teratas dengan Bintang Terbanyak di GitHub minggu ini, dimulai dari peringkat ke-10. Tautan ke repositori unggulan ada di komentar yang disematkan. 00:00 10 Teratas Minggu Ini 00:05 Peringkat ke-10 Android Skills 00:51 Peringkat ke-9 Orca 01:38 Peringkat ke-8 OpenCode 02:15 Peringkat ke-7 Luplo 02:47 Peringkat ke-6 Graphify 03:32 Peringkat ke-5 Trading Agents 04:17 Peringkat ke-4 Kapasi Skills 04:56 Peringkat ke-3 Claude SEO 05:35 Peringkat ke-2 Knowledge Work Plugin 06:14 Peringkat ke-1 Cloudflare Security Audit Skills 06:55 Ringkasan 📌 Minggu Ini! 10 Repositori Terpopuler di GitHub 10 Android Skills — android/skills 9 Orca — stablyai/orca 8 Open Code — anomalyco/opencode 7 Ruflo — ruvnet/ruflo 6 Graphify — Graphify-Labs/graphify 5 Trading Agents — TauricResearch/TradingAgents 4 Karpathy Skills — multica-ai/andrej-karpathy-skills 3 Claude SEO — AgriciDaniel/claude-seo 2 Knowledge Work Plugins — anthropics/knowledge-work-plugins 1 Cloudflare Security Audit Skills — cloudflare/security-audit-skill

@imwsl90 刷到一个skill有225k惊了😂 multica-ai/andrej-karpathy-skills

CLAUDE.md Best Practices: A Template for Unattended Agents Originally published at Unattended. CLAUDE.md best practices come down to three things. Keep the file short. Include only what Claude can't work out from the code. Write it so that a new agent with nobody around to ask can still finish the job. For unattended runs, that last point carries the most weight: the file should cover what to decide alone, what to stop on, and where to leave a note. If you only copy a style guide, the agent drifts, stalls, or keeps hitting permission denials overnight. Anthropic's own test for each line is: "Would removing this cause Claude to make mistakes?" Which option is right for you? Choose by how Claude will run, not by how big the project is. If you're unsure, start with the first row. | Your situation | What to pick first | First page to open | |---|---|---| | Beginner, no CLAUDE.md yet | Run /init, then cut the draft down to commands, rules, and "never" lines | Best practices: Write an effective CLAUDE.md | | Daily interactive work | A short project CLAUDE.md in git, plus skills for workflows you only need sometimes | Skills docs | | Nightly cron or claude -p job | The unattended template below, plus permission rules in settings | Headless mode | | CI with --bare | Pass the file yourself with --append-system-prompt-file | Bare mode section | | Tight token budget | Move rarely used content into skills or @ imports that only load when needed | /claude-code-how-to-see-token-usage-and-reduce-it/ | | Rules that must never break | A PreToolUse hook or a deny rule instead of a CLAUDE.md line | Hooks guide | Here's why, one line each. /init gives you a draft quickly, but the draft is too long. Skills keep everyday sessions lean. Unattended runs can't ask you anything. --bare doesn't load CLAUDE.md at all. Every line in CLAUDE.md costs tokens on every run. And a hook is the only kind of rule that can't be talked out of. What should a CLAUDE.md file example look like? A good CLAUDE.md example is short and scoped. It covers project context, commands, conventions, and explicit rules for what the agent must never do unattended. It isn't a full style guide. Anthropic's own sample is just two headings, "Code style" and "Workflow", with two bullets each. The docs say "There's no required format for CLAUDE.md files, but keep it short and human-readable." An unattended agent needs a few extra sections that an interactive user never thinks about, because the interactive user is the one answering questions. Here is an example for a scheduled job. It's illustrative, so replace the commands and paths with your own: # Project Static blog generator (Node 22). Content in /content, build output in /dist. # Commands - Install: npm ci - Test: npm test -- --run - Build: npm run build (must exit 0 before any commit) # Conventions - ES modules only. No new dependencies without a note in REPORT.md. # Autonomy: decide alone - Fix failing tests caused by your own change. - Retry a flaky network step once, then move on. # Autonomy: stop and report - Anything touching /migrations, .env, or CI config. - A task that needs more than ~10 files changed. - Any ambiguity you would normally ask about. # When blocked - Do not wait for an answer. Write the question and what you tried to BLOCKED.md, commit nothing, and end the run. # Done means - Tests pass, build exits 0, REPORT.md lists files changed and commands run. # Compaction - When compacting, always preserve the full list of modified files and any test commands. Each section maps to a documented behavior. "Done means" comes from Anthropic's advice to give Claude a check it can run. Their wording: it's "the difference between a session you watch and one you walk away from." The compaction line is close to word for word what the best practices page suggests for keeping critical context through summarization. The "When blocked" section exists because a run started with --permission-prompts none removes tools that need a person, such as AskUserQuestion, according to the headless docs. Without that section, Claude has no planned route when it wants to ask you something. Common pitfalls in examples: - Putting permissions in prose. "Never run rm -rf" in CLAUDE.md is a request, not a block. Use a deny rule or hook instead (see /claude-code-permissions-best-practices-headless-template/). - Describing the codebase file by file. Anthropic's exclude list names this directly, along with "Self-evident practices like 'write clean code'" (best practices). - No definition of done. Without one, "Claude stops when the work looks done." What are the core CLAUDE.md best practices? Keep CLAUDE.md concise, include only information that changes decisions, and write it as onboarding for an agent that can't ask questions. The reason is the context window. Anthropic says "Claude's context window fills up fast, and performance degrades as it fills," and CLAUDE.md loads at the start of every session. The memory docs add that Claude treats these files as "context, not enforced configuration." Every line costs tokens, and none of them is guaranteed to be followed. Anthropic's include/exclude table (source), rearranged for unattended use: | Include | Exclude | |---|---| | Bash commands Claude can't guess | Anything Claude can figure out by reading code | | Code style rules that differ from defaults | Standard language conventions | | Test instructions and preferred runner | Detailed API docs (link instead) | | Branch naming and PR conventions | Information that changes frequently | | Project-specific architecture decisions | Long explanations or tutorials | | Required env vars and environment quirks | File-by-file descriptions | | Non-obvious gotchas | "Write clean code" | Steps to write or rewrite yours: 1. In the project folder, run claude, then type /init to generate a starter file (best practices). 2. Delete every line that fails the test "Would removing this cause Claude to make mistakes?" 3. Add the autonomy, "when blocked", and "done means" sections from the example above. 4. Run /context to confirm the file loaded (best practices). 5. For a checked-in file, run /doctor, and Claude proposes cuts for content it can work out from the codebase (best practices). 6. Commit it to git. Two failure signals tell you which way to fix the file. Anthropic puts it this way: "If Claude keeps doing something you don't want despite having a rule against it, the file is probably too long." If Claude asks about something the file already answers, the wording is probably ambiguous (best practices). Use emphasis sparingly. Mark only the one rule Claude keeps skipping with "IMPORTANT", because "If you emphasize many lines, none of them stands out." What CLAUDE.md patterns do top GitHub repos use? Popular GitHub repos usually use three CLAUDE.md patterns: a command-and-convention sheet, a behavior-principles file, or a shared team file updated after every mistake. | Pattern | Public example | What it contains | Good for unattended runs? | |---|---|---|---| | Command and convention sheet | EDSL file in awesome-claude-code | Build/test commands, strict style (Black, mypy), testing requirements | Yes, if you add stop rules | | Behavior principles | multica-ai/andrej-karpathy-skills | Four principles about assumptions, simplicity, edit scope, goals | Partly; it has no project commands | | Living team file | Claude Code team's own file (Boris Cherny on X) | Corrections added whenever Claude gets something wrong | Yes; it grows from real failures | A few structural habits show up again and again. One is splitting rules by directory or file type: Anthropic documents nested CLAUDE.md files and a .claude/rules/ folder for rules scoped to specific files (memory docs). Another is pointing to longer docs with @path/to/import instead of pasting them in (best practices). Imports help with organization. They don't make the file cheaper if the imported file loads every session anyway. If your repo already has an AGENTS.md, the memory docs say Claude "can also read a repository's AGENTS.md files in place of CLAUDE.md," so you don't need two copies. Limitation: most public examples are written for someone sitting at the terminal. Hardly any say what the agent should do when it's stuck at 3 a.m., and that's the section you'll need to write yourself. What does Anthropic's official guidance say about CLAUDE.md? Anthropic's official guidance treats CLAUDE.md as a short, advisory memory file that loads every session. It covers commands, conventions, and gotchas, and you refine it like code. The points that matter most for automation, all from Anthropic's docs as of October 2026: - Scopes. CLAUDE.md can live at managed-policy, user (~/.claude/CLAUDE.md), project (./CLAUDE.md or ./.claude/CLAUDE.md), and local (./CLAUDE.local.md) levels (memory docs). On a VPS, the user-level file belongs to whatever account the cron job runs as. That may not be the account you set up interactively. - Auto memory is separate. Claude writes its own notes, and only the first 200 lines or 25KB load each session (memory docs). - Hooks vs instructions. "Unlike CLAUDE.md instructions which are advisory, hooks are deterministic and guarantee the action happens". To block an action no matter what Claude decides, the memory docs point to a PreToolUse hook (memory docs). - Treat it like code. "Review it when things go wrong, prune it regularly, and test changes by observing whether Claude's behavior actually shifts" (best practices). The headless page is where unattended setups most often go wrong. Adding --bare skips auto-discovery of "hooks, skills, custom commands, subagents, installed plugins, MCP servers, auto memory, and CLAUDE.md" (headless docs). The same page warns that "--bare is the recommended mode for scripted and SDK calls, and will become the default for -p in a future release." So a cron job that depends on CLAUDE.md being picked up automatically may stop following it after an update, with no error to tell you. To make the file load whatever the default turns out to be: 1. Keep the unattended rules in a file, for example CLAUDE.md or a separate AGENT_RULES.md. 2. Pass it explicitly. The headless docs list --append-system-prompt-file for adding to the system prompt in bare mode: claude --bare -p "Run the nightly content check" \ --append-system-prompt-file ./CLAUDE.md \ --allowedTools "Read,Edit,Bash(npm test *)" \ --permission-prompts none 3. Remember that bare mode doesn't use your subscription login. Set ANTHROPIC_API_KEY first (headless docs). Two more defaults matter. With Claude Code v2.1.302 or later, claude -p starts in Manual mode when nothing sets a permission mode, and a call that needs approval is denied unless something approves it. --permission-prompts requires v2.1.259 or later (headless docs). This is why CLAUDE.md has to tell the agent what to do after a denial. The full flag set is covered in /claude-code-headless-mode-flags-auth-and-cost-2026/. What do Reddit discussions say works (and fails) in CLAUDE.md? The usual complaint on Reddit and developer forums is that long CLAUDE.md files get ignored. The fix people recommend is to trim the file to essentials and prune it regularly. Plenty of community posts carry titles like "Why your long CLAUDE.md gets ignored, and how to fix it", and forum threads ask whether popular rule files actually help (Devtalk). Treat these as anecdotes, not measurements. The pattern does line up with Anthropic's own failure list, which names "The over-specified CLAUDE.md" and says important rules "get lost in the noise". What community advice tends to get right: - Add rules only after a real mistake. That matches the practice used by the Claude Code team (next sections). - Turn rules into hooks. Anthropic's own fix for an over-specified file: "If Claude already does something correctly without the instruction, delete it or convert it to a hook." What it tends to get wrong for unattended work: - Shouting. Files full of "ALWAYS" and "NEVER" in capitals cancel themselves out, since emphasis only works on one line. - Language and tone rules at the top. In a nightly job nobody reads the tone. Spend those lines on stop conditions instead. For example, say your overnight agent keeps editing CI config even though a rule forbids it. Community advice would tell you to move the rule higher in the file. The more reliable fix is a deny rule or a PreToolUse hook on that path, with one line left in CLAUDE.md explaining why. Copy-paste hook scripts for this are in /claude-code-hooks-examples-7-copy-paste-scripts/. What does Andrej Karpathy say about CLAUDE.md style files? Karpathy didn't write a CLAUDE.md himself. The popular "Karpathy CLAUDE.md" is a community file built from his posts on how LLMs fail at coding, and it focuses on behavior rather than style. The repo multica-ai/andrej-karpathy-skills showed about 223.8k stars when checked in October 2026. It turns his observations into four principles: Think Before Coding, Simplicity First, Surgical Changes, and Goal-Driven Execution. The line that matters most for unattended runs is the one the repo quotes from Karpathy: "Don't tell it what to do, give it success criteria and watch it go." How to use it in a scheduled setup: 1. Keep "Surgical Changes" and "Goal-Driven Execution". They limit drift and define when the agent is finished. 2. Rewrite "Think Before Coding" for unattended runs. The original tells the agent to ask for clarification when a request is ambiguous. Overnight, nobody is there to answer, so change that to "write the question to BLOCKED.md and stop." 3. Add your own commands. The file has no project-specific build or test lines. Who it's not for: if your CLAUDE.md is already near the size where rules get lost, pasting in a generic principles file makes that worse. Take the two principles you need and leave the rest. How is CLAUDE.md best practice evolving for 2026 (skills, headless agents)? In 2026, CLAUDE.md is shrinking. Workflows are moving into skills, hard rules into hooks and permission settings, and the core file is turning into a statement of autonomy boundaries. The split, according to Anthropic's docs as of October 2026: | Content | Put it in | Why | |---|---|---| | Always-true commands, conventions, stop rules | CLAUDE.md | "CLAUDE.md is loaded every session, so only include things that apply broadly" | | Occasional domain knowledge or workflows | .claude/skills/<name>/SKILL.md | Claude loads skills on demand (best practices) | | Workflows with side effects | Skill with disable-model-invocation: true | Only runs when you trigger it (best practices) | | Must-never-happen actions | Hooks, deny rules | Deterministic (hooks guide) | | "Keep going until tests pass" | /goal condition or a Stop hook | Lets "an unattended run finish correctly without you" | Skills also work headless. User-invoked skills run in -p mode if you include /skill-name in the prompt string (headless docs). A clean nightly setup looks like this: 1. CLAUDE.md, about a screen long: commands, autonomy rules, "when blocked", "done means". 2. A skill such as /nightly-report that holds the step-by-step procedure. 3. Cron calls claude -p "/nightly-report" --permission-mode dontAsk --allowedTools "...". Under dontAsk, Claude Code "denies every call that would otherwise prompt" (headless docs). 4. A Stop hook runs your test command, so the run can't end while tests are red. Auto mode is the other 2026 change. From v2.1.283 it's the starting permission mode for interactive terminal and VS Code sessions, with a classifier that blocks actions like scope escalation (best practices). It reduces prompts, but it doesn't replace stop rules. The classifier judges risk. It doesn't know your business rules, such as "never publish on Sundays". Scheduling details are in /claude-code-cron-job-setup-run-automations-overnight/. What do practitioner write-ups (Medium, Boris Cherny) add to best practices? Practitioners, including Claude Code creator Boris Cherny, add one habit: update CLAUDE.md after every mistake you catch. Keep the file small enough that the new rule doesn't get lost. Cherny described the Claude Code team's setup on X in January 2026. They share one CLAUDE.md for the repo, check it into git, and "the whole team contributes multiple times a week". Whenever Claude gets something wrong, they add it to the file. In a follow-up post, he said he tags @.claude on coworkers' PRs to add CLAUDE.md lines through the Claude Code GitHub Action. For very long tasks, he mentioned having Claude verify its work or using a Stop hook (thread). Adapted for unattended runs: 1. Each morning, read the run's REPORT.md and BLOCKED.md. 2. For each real mistake, decide on the fix. A one-line CLAUDE.md rule works if the mistake was a judgment call. A hook or deny rule is better if it must never happen again. 3. Add the line in the same commit as the fix. 4. Once a month, delete rules the agent now follows without being told. Pitfall: a file that only grows. Adding a line after every mistake without ever pruning produces exactly the bloated file Anthropic warns about. A related mistake is reacting to a stalled run by adding more instructions. Stalls usually come from permissions or context, not missing rules. Check /why-claude-code-stops-mid-automation-and-how-to-fix-it/ before you edit the file. Next steps Do today (DIY): 1. Open your project, run /init (or open your existing file with /memory), and cut every line that fails "Would removing this cause Claude to make mistakes?" 2. Add four sections from the example: "Autonomy: decide alone", "Autonomy: stop and report", "When blocked", and "Done means". 3. If your cron job uses -p, add --append-system-prompt-file ./CLAUDE.md now, so the coming --bare default doesn't silently drop the file. If you'd rather not do it yourself: Anthropic's paid plans include Claude Code. Pro is $20/month ($17/month billed annually), and Max starts at $100/month (as of October 2026; check the official pricing page). Pro suits one person running light scheduled jobs. It's not a good fit if you need many long runs per night, because Claude Code "shares the same usage limits as the rest of your plan." Bare-mode runs bill through an API key instead, so they don't draw on a subscription. Anthropic also publishes a Claude Code 101 course on CLAUDE.md; check the page for current access terms. It suits beginners, but it won't write your stop rules for you. Related guides on this site: - /claude-code-permissions-best-practices-headless-template/ - /claude-code-hooks-examples-7-copy-paste-scripts/ - /claude-code-headless-mode-flags-auth-and-cost-2026/ - /claude-code-cron-job-setup-run-automations-overnight/ - /why-claude-code-stops-mid-automation-and-how-to-fix-it/ FAQ Where can I find CLAUDE.md best practices examples? Anthropic's best practices page has a short official sample and an include/exclude table. Community examples are collected in awesome-claude-code. The template in this guide adds the unattended sections most public examples leave out. What are Anthropic's CLAUDE.md best practices? Keep it short. Include only commands, rules that differ from defaults, and gotchas. Check it into git and prune it regularly. Anthropic also says CLAUDE.md is advisory, so anything that must always happen belongs in a hook (best practices). Are there good CLAUDE.md best practices on GitHub? Yes. The most-starred example is multica-ai/andrej-karpathy-skills, a behavior-principles file. Language-specific files such as the EDSL example in awesome-claude-code show the command-sheet pattern. Either way, add your own commands and stop rules. What are the Karpathy CLAUDE.md best practices? They're four principles taken from Karpathy's posts: Think Before Coding, Simplicity First, Surgical Changes, and Goal-Driven Execution. Karpathy didn't publish the file himself; the multica-ai repo did. For unattended runs, change "ask for clarification" to "log the question and stop". Sources - Best practices for Claude Code (Anthropic) - How Claude remembers your project (Anthropic memory docs) - Run Claude Code programmatically (headless docs) - Hooks guide (Anthropic) - Skills (Anthropic) - Claude pricing - Claude Code 101: The CLAUDE.md file (Anthropic Academy) - Boris Cherny on X: shared CLAUDE.md - Boris Cherny on X: tagging @.claude in PRs - Boris Cherny on X: long-running tasks - multica-ai/andrej-karpathy-skills (GitHub) - awesome-claude-code EDSL CLAUDE.md (GitHub) - Why your long CLAUDE.md gets ignored (DEV Community) - Has anyone tried the Karpathy CLAUDE.md rules? (Devtalk) - Writing the Best CLAUDE.md (DataCamp) - Creating the Perfect CLAUDE.md (Dometrain) --- This article was created with the help of AI. Sources are linked inline; check the official documentation for the latest details.

Top 10 GitHub Repos This Week | Oct 18 – Oct 24 | The Harness Beats the Model This week the biggest gains went to the harness around the agent, not the model: Cloudflare's security audit skill took #1 with 12,680 new stars, and three of the ten are instructions rather than apps. ⏱️ TIMESTAMPS 0:00 — Intro 0:46 — #10 andrej-karpathy-skills — One rules file for coding agents (223.6K⭐️) 2:37 — #9 claude-plugins-official — Anthropic's official plugin directory (44.5K⭐️) 4:36 — #8 ruflo — An agent meta-harness (79.2K⭐️) 6:38 — #7 ai-memory — Shared memory across coding agents (13.4K⭐️) 8:46 — #6 ComfyUI — Node-graph engine for AI images and video (142.8K⭐️) 10:50 — #5 hyperframes — Write HTML, render video (71.9K⭐️) 13:13 — #4 opencode — The open source coding agent (221.4K⭐️) 15:21 — #3 graphify — Your codebase as a knowledge graph (134.8K⭐️) 18:01 — #2 orca — A command center for a fleet of agents (105.4K⭐️) 20:18 — #1 security-audit-skill — Turn your agent into a security auditor (46K⭐️) 23:17 — Recap & Trends 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Like, 💬 Comment, 🔔 Subscribe, ⭐️ Star the repos! #github #opensource #ai #coding #programming #trending #devtools #developer [ar] أفضل 10 مستودعات على GitHub لهذا الأسبوع | 18 أكتوبر – 24 أكتوبر | "إطار العمل" يتفوق على النموذج هذا الأسبوع، كانت أكبر المكاسب من نصيب البنية التحتية المحيطة بالوكيل، وليس النموذج نفسه: فقد تصدّرت مهارة التدقيق الأمني ​​من كلاود فلير القائمة بحصولها على 12,680 نجمة جديدة، وثلاث من هذه النجوم العشر عبارة عن تعليمات وليست تطبيقات. ⏱️ الطوابع الزمنية 0:00 — مقدمة 0:46 — #10 مهارات أندريه كارباثي — ملف قواعد واحد لوكلاء البرمجة (223.6 ألف نجمة) 2:37 — #9 دليل إضافات كلود الرسمي — دليل إضافات أنثروبيك الرسمي (44.5 ألف نجمة) 4:36 — #8 روفلو — أداة شاملة لوكلاء البرمجة (79.2 ألف نجمة) 6:38 — #7 ذاكرة الذكاء الاصطناعي — ذاكرة مشتركة بين وكلاء البرمجة (13.4 ألف نجمة) 8:46 — #6 كومفي يو آي — محرك رسم بياني للعقد لصور وفيديوهات الذكاء الاصطناعي (142.8 ألف نجمة) 10:50 — #5 هايبر فريمز — كتابة HTML، عرض الفيديو (71.9 ألف نجمة) 13:13 — #4 أوبن كود — وكيل البرمجة مفتوح المصدر (221.4 ألف نجمة) 15:21 — #3 graphify — قاعدة بياناتك البرمجية كرسم بياني معرفي (134.8 ألف نجمة) 18:01 — #2 orca — مركز قيادة لأسطول من العملاء (105.4 ألف نجمة) 20:18 — #1 security-audit-skill — حوّل عميلك إلى مدقق أمني (46 ألف نجمة) 23:17 — ملخص واتجاهات 🔗 المستودعات • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → staablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 إعجاب، 💬 تعليق، 🔔 اشتراك، ⭐️ ضع نجمة على المستودعات! #github #opensource #ai #coding #programming #trending #devtools #developer [bn] এই সপ্তাহের শীর্ষ ১০ গিটহাব রিপো | ১৮ অক্টোবর – ২৪ অক্টোবর | মডেলকে হার মানালো হারনেস এই সপ্তাহে সবচেয়ে বেশি অগ্রগতি হয়েছে এজেন্টের চারপাশের হারনেসের, মডেলের নয়: ক্লাউডফ্লেয়ারের সিকিউরিটি অডিট স্কিল ১২,৬৮০টি নতুন স্টার পেয়ে প্রথম স্থান অধিকার করেছে, এবং এই দশটির মধ্যে তিনটিই অ্যাপ না হয়ে নির্দেশনা। ⏱️ টাইমস্ট্যাম্প ০:০০ — ভূমিকা ০:৪৬ — #১০ andrej-karpathy-skills — কোডিং এজেন্টদের জন্য একটি রুলস ফাইল (২২৩.৬ হাজার⭐️) ২:৩৭ — #৯ claude-plugins-official — অ্যানথ্রোপিকের অফিসিয়াল প্লাগইন ডিরেক্টরি (৪৪.৫ হাজার⭐️) ৪:৩৬ — #৮ ruflo — একটি এজেন্ট মেটা-হারনেস (৭৯.২ হাজার⭐️) ৬:৩৮ — #৭ ai-memory — কোডিং এজেন্টদের মধ্যে শেয়ার্ড মেমরি (১৩.৪ হাজার⭐️) ৮:৪৬ — #৬ ComfyUI — AI ইমেজ এবং ভিডিওর জন্য নোড-গ্রাফ ইঞ্জিন (১৪২.৮ হাজার⭐️) ১০:৫০ — #৫ hyperframes — HTML লেখা, ভিডিও রেন্ডার করা (৭১.৯ হাজার⭐️) ১৩:১৩ — #৪ opencode — ওপেন সোর্স কোডিং এজেন্ট (২২১.৪ হাজার ⭐️) ১৫:২১ — #৩ গ্রাফিফাই — আপনার কোডবেসকে একটি নলেজ গ্রাফ হিসেবে ব্যবহার করুন (১৩৪.৮ হাজার ⭐️) ১৮:০১ — #২ ওরকা — একদল এজেন্টের জন্য একটি কমান্ড সেন্টার (১০৫.৪ হাজার ⭐️) ২০:১৮ — #১ সিকিউরিটি-অডিট-স্কিল — আপনার এজেন্টকে একজন সিকিউরিটি অডিটরে পরিণত করুন (৪৬ হাজার ⭐️) ২৩:১৭ — সারসংক্ষেপ ও ট্রেন্ডস 🔗 রিপো • #১০ আন্দ্রেজ-কারপ্যাথি-স্কিলস → multica-ai/andrej-karpathy-skills • #৯ ক্লড-প্লাগইনস-অফিশিয়াল → anthropics/claude-plugins-official • #৮ রুফলো → ruvnet/ruflo • #৭ ai-memory → akitaonrails/ai-memory • #৬ ComfyUI → Comfy-Org/ComfyUI • #৫ hyperframes → heygen-com/hyperframes • #৪ opencode → anomalyco/opencode • #৩ graphify → Graphify-Labs/graphify • #২ orca → stablyai/orca • #১ security-audit-skill → cloudflare/security-audit-skill 👍 লাইক, 💬 কমেন্ট, 🔔 সাবস্ক্রাইব, ⭐️ রিপোজিটরিগুলোতে স্টার দিন! #github #opensource #ai #coding #programming #trending #devtools #developer [de-DE] Die 10 besten GitHub-Repos der Woche | 18. Okt. – 24. Okt. | Das Framework schlägt das Modell Diese Woche erzielten die Anwender die größten Zuwächse, nicht das Modell selbst: Cloudflares Sicherheitsaudit-Funktion belegte mit 12.680 neuen Sternen den ersten Platz, und drei der zehn besten Funktionen sind Anleitungen und keine Apps. ⏱️ ZEITSTEMPEL 0:00 — Intro 0:46 — #10 andrej-karpathy-skills — Eine Regeldatei für Coding-Agenten (223,6K⭐️) 2:37 — #9 claude-plugins-official — Das offizielle Plugin-Verzeichnis von Anthropic (44,5K⭐️) 4:36 — #8 ruflo — Ein Agenten-Meta-Harness (79,2K⭐️) 6:38 — #7 ai-memory — Gemeinsamer Speicher für Coding-Agenten (13,4K⭐️) 8:46 — #6 ComfyUI — Node-Graph-Engine für KI-Bilder und -Videos (142,8K⭐️) 10:50 — #5 hyperframes — HTML schreiben, Video rendern (71,9K⭐️) 13:13 — #4 Opencode – Der Open-Source-Codierungsagent (221,4K⭐️) 15:21 – #3 Graphify – Ihre Codebasis als Wissensgraph (134,8K⭐️) 18:01 – #2 Orca – Eine Kommandozentrale für eine Flotte von Agenten (105,4K⭐️) 20:18 – #1 Security-Audit-Skill – Machen Sie Ihren Agenten zum Sicherheitsauditor (46K⭐️) 23:17 – Zusammenfassung & Trends 🔗 REPOS • #10 Andrej-Karpathy-Skills → multica-ai/andrej-karpathy-skills • #9 Claude-Plugins-Official → anthropics/claude-plugins-official • #8 Ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Gefällt mir, 💬 Kommentar, 🔔 Abonnieren, ⭐️ Repos mit Stern markieren! #github #opensource #ai #coding #programming #trending #devtools #developer [es-US] Los 10 mejores repositorios de GitHub esta semana | 18 al 24 de octubre | The Harness Beats the M... Esta semana, las mayores ganancias se concentraron en la plataforma que rodea al agente, no en el modelo: la habilidad de auditoría de seguridad de Cloudflare alcanzó el primer puesto con 12.680 nuevas estrellas, y tres de las diez mejores son instrucciones, no aplicaciones. Esta semana, el mayor avance se centró en la plataforma que rodea al agente, no en el modelo: la habilidad de auditoría de seguridad de Cloudflare se posicionó en el primer lugar con 12.680 nuevas estrellas, y tres de las diez mejores son instrucciones, no aplicaciones. ⏱️ MARCAS DE TIEMPO 0:00 — Introducción 0:46 — #10 andrej-karpathy-skills — Un archivo de reglas para agentes de codificación (223.6K⭐️) 2:37 — #9 claude-plugins-official — Directorio oficial de plugins de Anthropic (44.5K⭐️) 4:36 — #8 ruflo — Un meta-arnés para agentes (79.2K⭐️) 6:38 — #7 ai-memory — Memoria compartida entre agentes de codificación (13.4K⭐️) 8:46 — #6 ComfyUI — Motor de grafo de nodos para imágenes y vídeo con IA (142.8K⭐️) 10:50 — #5 hyperframes — Escribe HTML, renderiza vídeo (71.9K⭐️) 13:13 — #4 opencode — OpenCode Agente de codificación fuente (221.4K⭐️) 15:21 — #3 graphify — Tu base de código como un grafo de conocimiento (134.8K⭐️) 18:01 — #2 orca — Un centro de mando para una flota de agentes (105.4K⭐️) 20:18 — #1 security-audit-skill — Convierte a tu agente en un auditor de seguridad (46K⭐️) 23:17 — Resumen y tendencias 🔗 REPOSICIONES • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Dale Me gusta, 💬 Comenta, 🔔 Suscríbete, ⭐️ ¡Marca los repositorios como favoritos! #github #opensource #ai #codificación #programación #tendencias #devtools #desarrollador [fr-FR] Top 10 des dépôts GitHub de la semaine | 18-24 oct. | Le harnais surpasse le modèle Cette semaine, ce sont les outils autour de l'agent, et non le modèle, qui ont enregistré les plus fortes progressions : la fonctionnalité d'audit de sécurité de Cloudflare s'est hissée à la première place avec 12 680 nouvelles étoiles, et trois des dix meilleures sont des instructions plutôt que des applications. ⏱️ HORODATAGE 0:00 — Introduction 0:46 — #10 andrej-karpathy-skills — Un fichier de règles pour la programmation d'agents (223,6K⭐️) 2:37 — #9 claude-plugins-official — Répertoire officiel des plugins d'Anthropic (44,5K⭐️) 4:36 — #8 ruflo — Un méta-harnais pour agents (79,2K⭐️) 6:38 — #7 ai-memory — Mémoire partagée entre agents de programmation (13,4K⭐️) 8:46 — #6 ComfyUI — Moteur de graphes de nœuds pour les images et vidéos IA (142,8K⭐️) 10:50 — #5 hyperframes — Écrire du HTML, générer de la vidéo (71,9K⭐️) 13:13 — #4 opencode — L'agent de codage open source (221,4K⭐️) 15:21 — #3 Graphify — Votre code source sous forme de graphe de connaissances (134,8K⭐️) 18:01 — #2 Orca — Un centre de commandement pour une flotte d'agents (105,4K⭐️) 20:18 — #1 Security-audit-skill — Transformez votre agent en auditeur de sécurité (46K⭐️) 23:17 — Récapitulatif et tendances 🔗 DÉPÔTS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Aimez, 💬 Commentez, 🔔 Abonnez-vous, ⭐️ Mettez les dépôts en favoris ! #github #opensource #ai #coding #programming #trending #devtools #developer [hi] इस सप्ताह के शीर्ष 10 GitHub रिपोज़ | 18 अक्टूबर - 24 अक्टूबर | हार्नेस मॉडल से बेहतर इस सप्ताह सबसे अधिक लाभ एजेंट के आसपास के हार्नेस को मिला, न कि मॉडल को: क्लाउडफ्लेयर के सुरक्षा ऑडिट कौशल ने 12,680 नए स्टार्स के साथ पहला स्थान प्राप्त किया, और इनमें से दस में से तीन निर्देश हैं, न कि ऐप्स। ⏱️ समय-सारणी 0:00 — परिचय 0:46 — #10 andrej-karpathy-skills — कोडिंग एजेंटों के लिए एक नियम फ़ाइल (223.6K⭐️) 2:37 — #9 claude-plugins-official — Anthropic की आधिकारिक प्लगइन निर्देशिका (44.5K⭐️) 4:36 — #8 ruflo — एक एजेंट मेटा-हार्नेस (79.2K⭐️) 6:38 — #7 ai-memory — कोडिंग एजेंटों के बीच साझा मेमोरी (13.4K⭐️) 8:46 — #6 ComfyUI — AI छवियों और वीडियो के लिए नोड-ग्राफ़ इंजन (142.8K⭐️) 10:50 — #5 hyperframes — HTML लिखें, वीडियो रेंडर करें (71.9K⭐️) 13:13 — #4 opencode — ओपन सोर्स कोडिंग एजेंट (221.4K⭐️) 15:21 — #3 ग्राफीफाई — आपका कोडबेस एक नॉलेज ग्राफ के रूप में (134.8K⭐️) 18:01 — #2 ओर्का — एजेंटों के बेड़े के लिए एक कमांड सेंटर (105.4K⭐️) 20:18 — #1 सिक्योरिटी ऑडिट स्किल — अपने एजेंट को सिक्योरिटी ऑडिटर में बदलें (46K⭐️) 23:17 — सारांश और रुझान 🔗 रिपॉजिटरी • #10 आंद्रेज कार्पेथी स्किल्स → multica-ai/andrej-karpathy-skills • #9 क्लाउड प्लगइन्स ऑफिशियल → anthropics/claude-plugins-official • #8 रुफ्लो → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 लाइक करें, 💬 टिप्पणी करें, 🔔 सब्सक्राइब करें, ⭐️ रिपॉजिटरी को स्टार दें! #github #opensource #ai #coding #programming #trending #devtools #developer [id] 10 Repositori GitHub Teratas Minggu Ini | 18 Okt – 24 Okt | The Harness Beats the Model Minggu ini, peningkatan terbesar terjadi pada sistem pendukung agen, bukan pada modelnya: keterampilan audit keamanan Cloudflare menduduki peringkat #1 dengan 12.680 bintang baru, dan tiga dari sepuluh bintang tersebut berupa instruksi, bukan aplikasi. ⏱️ CAP WAKTU 0:00 — Pendahuluan 0:46 — #10 andrej-karpathy-skills — Satu file aturan untuk agen pengkodean (223,6K⭐️) 2:37 — #9 claude-plugins-official — Direktori plugin resmi Anthropic (44,5K⭐️) 4:36 — #8 ruflo — Meta-harness agen (79,2K⭐️) 6:38 — #7 ai-memory — Memori bersama di seluruh agen pengkodean (13,4K⭐️) 8:46 — #6 ComfyUI — Mesin node-graph untuk gambar dan video AI (142,8K⭐️) 10:50 — #5 hyperframes — Tulis HTML, render video (71,9K⭐️) 13:13 — #4 opencode — Agen pengkodean sumber terbuka (221,4 ribu ⭐️) 15:21 — #3 graphify — Basis kode Anda sebagai grafik pengetahuan (134,8 ribu ⭐️) 18:01 — #2 orca — Pusat komando untuk armada agen (105,4 ribu ⭐️) 20:18 — #1 security-audit-skill — Ubah agen Anda menjadi auditor keamanan (46 ribu ⭐️) 23:17 — Rekap & Tren 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Suka, 💬 Komentar, 🔔 Berlangganan, ⭐️ Beri bintang pada repositori! #github #opensource #ai #coding #programming #trending #devtools #developer [it] I 10 migliori repository GitHub di questa settimana | 18 ott – 24 ott | L'infrastruttura batte il... Questa settimana i maggiori progressi sono andati all'infrastruttura che circonda l'agente, non al modello: la skill di audit di sicurezza di Cloudflare si è aggiudicata il primo posto con 12.680 nuove stelle, e tre delle prime dieci sono istruzioni anziché applicazioni. ⏱️ INDICE DEI CONTENUTI 0:00 — Introduzione 0:46 — #10 andrej-karpathy-skills — Un file di regole per agenti di codifica (223.600⭐️) 2:37 — #9 claude-plugins-official — La directory ufficiale dei plugin di Anthropic (44.500⭐️) 4:36 — #8 ruflo — Un meta-sistema di controllo per agenti (79.200⭐️) 6:38 — #7 ai-memory — Memoria condivisa tra agenti di codifica (13.400⭐️) 8:46 — #6 ComfyUI — Motore a grafo di nodi per immagini e video AI (142.800⭐️) 10:50 — #5 hyperframes — Scrivi HTML, renderizza video (71.900⭐️) 13:13 — #4 opencode — Il progetto open source agente di codifica (221.400⭐️) 15:21 — #3 graphify — Il tuo codice sorgente come grafo della conoscenza (134.800⭐️) 18:01 — #2 orca — Un centro di comando per una flotta di agenti (105.400⭐️) 20:18 — #1 security-audit-skill — Trasforma il tuo agente in un revisore della sicurezza (46.000⭐️) 23:17 — Riepilogo e tendenze 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Metti Mi piace, 💬 Commenta, 🔔 Iscriviti, ⭐️ Aggiungi i repository ai preferiti! #github #opensource #ai #coding #programming #trending #devtools #developer [iw] 10 מאגרי ה-GitHub המובילים השבוע | 18 באוקטובר – 24 באוקטובר | המעטפת מנצחת את המודל השבוע, הרווחים הגדולים ביותר הלכו לרתמה סביב הסוכן, לא למודל: מיומנות ביקורת האבטחה של Cloudflare תפסה את המקום הראשון עם 12,680 כוכבים חדשים, ושלושה מתוך עשרתם הם הוראות ולא אפליקציות. ⏱️ חותמות זמן 0:00 — מבוא 0:46 — #10 andrej-karpathy-skills — קובץ כללים אחד עבור סוכני קידוד (223.6K⭐️) 2:37 — #9 claude-plugins-official — ספריית התוספים הרשמית של Anthropic (44.5K⭐️) 4:36 — #8 ruflo — רתמת מטא של סוכן (79.2K⭐️) 6:38 — #7 ai-memory — זיכרון משותף בין סוכני קידוד (13.4K⭐️) 8:46 — #6 ComfyUI — מנוע Node-graph עבור תמונות ווידאו של בינה מלאכותית (142.8K⭐️) 10:50 — #5 hyperframes — כתיבת HTML, עיבוד וידאו (71.9K⭐️) 13:13 — #4 opencode — סוכן קידוד בקוד פתוח (221.4K⭐️) 15:21 — #3 graphify — בסיס הקוד שלך כגרף ידע (134.8K⭐️) 18:01 — #2 orca — מרכז פיקוד עבור צי סוכנים (105.4K⭐️) 20:18 — #1 security-audit-skill — הפוך את הסוכן שלך למבקר אבטחה (46K⭐️) 23:17 — סיכום ומגמות 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 היפר-פריימים → heygen-com/hyperframes • #4 קוד פתוח → anomalyco/opencode • #3 גרפיפי → Graphify-Labs/graphify • #2 אורקה → stablyai/orca • #1 סקיל ביקורת-על-ביטחון → cloudflare/security-audit-skill 👍 לייק, 💬 הגיבו, 🔔 הירשמו, ⭐️ סמנו בכוכב את המאגרים! #github #קודפתוח #בינה מלאכותית #קידוד #תכנות #טרנדי #כליפיתוח #מפתח [ja] 今週のGitHub注目リポジトリTOP10 | 10月18日～10月24日 | モデルを凌駕するハーネス 今週、最も大きな伸びを見せたのはエージェントのモデルではなく、その周辺技術でした。Cloudflareのセキュリティ監査スキルが12,680個の新たなスターを獲得し、1位に輝きました。上位10件のうち3件はアプリではなく、手順書でした。 ⏱️ タイムスタンプ 0:00 — はじめに 0:46 — #10 andrej-karpathy-skills — コーディングエージェントのためのルールファイル (223.6K⭐️) 2:37 — #9 claude-plugins-official — Anthropic の公式プラグインディレクトリ (44.5K⭐️) 4:36 — #8 ruflo — エージェントメタハーネス (79.2K⭐️) 6:38 — #7 ai-memory — コーディングエージェント間で共有されるメモリ (13.4K⭐️) 8:46 — #6 ComfyUI — AI 画像と動画のためのノードグラフエンジン (142.8K⭐️) 10:50 — #5 hyperframes — HTML の書き込み、動画のレンダリング (71.9K⭐️) 13:13 — #4 opencode — オープンソースのコーディングエージェント (221.4K⭐️) 15:21 — #3 graphify — コードベースをナレッジグラフとして活用 (134.8K⭐️) 18:01 — #2 orca — エージェント群のためのコマンドセンター (105.4K⭐️) 20:18 — #1 security-audit-skill — エージェントをセキュリティ監査ツールに変える (46K⭐️) 23:17 — まとめとトレンド 🔗 リポジトリ • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 ハイパーフレーム → heygen-com/hyperframes • #4 オープンコード → anomalyco/opencode • #3 グラフィファイ → Graphify-Labs/graphify • #2 オルカ → stablyai/orca • #1 セキュリティ監査スキル → cloudflare/security-audit-skill 👍 いいね、💬 コメント、🔔 チャンネル登録、⭐️ スターをリポジトリに追加！ #github #opensource #ai #coding #programming #trending #devtools #developer [ko] 이번 주 GitHub Top 10 저장소 | 10월 18일 – 10월 24일 | 모델을 능가하는 활용 도구 이번 주 가장 큰 성과는 모델이 아닌 에이전트 관련 도구에서 나타났습니다. 클라우드플레어의 보안 감사 스킬이 12,680개의 새로운 별점을 얻으며 1위를 차지했고, 상위 10개 중 3개는 앱이 아닌 사용법 안내였습니다. ⏱️ 타임스탬프 0:00 — 소개 0:46 — #10 andrej-karpathy-skills — 코딩 에이전트를 위한 하나의 규칙 파일 (223.6K⭐️) 2:37 — #9 claude-plugins-official — Anthropic의 공식 플러그인 디렉토리 (44.5K⭐️) 4:36 — #8 ruflo — 에이전트 메타 하네스 (79.2K⭐️) 6:38 — #7 ai-memory — 코딩 에이전트 간 공유 메모리 (13.4K⭐️) 8:46 — #6 ComfyUI — AI 이미지 및 비디오용 노드 그래프 엔진 (142.8K⭐️) 10:50 — #5 hyperframes — HTML 작성 및 비디오 렌더링 (71.9K⭐️) 13:13 — #4 opencode — 오픈 소스 코딩 에이전트 (221.4K⭐️) 15:21 — #3 Graphify — 코드베이스를 지식 그래프로 (134.8K⭐️) 18:01 — #2 Orca — 에이전트 함대를 위한 명령 센터 (105.4K⭐️) 20:18 — #1 Security-Audit-Skill — 에이전트를 보안 감사자로 전환 (46K⭐️) 23:17 — 요약 및 트렌드 🔗 저장소 • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 좋아요, 💬 댓글, 🔔 구독, ⭐️ 저장소에 별표 표시! #github #opensource #ai #coding #programming #trending #devtools #developer [ml] ഈ ആഴ്ചയിലെ മികച്ച 10 GitHub റിപ്പോസിറ്ററികൾ | ഒക്ടോബർ 18 – ഒക്ടോബർ 24 | മോഡലിനേക്കാൾ കേമൻ ഹാർനെസ് ഈ ആഴ്ച ഏറ്റവും വലിയ നേട്ടം മോഡലിനെയല്ല, ഏജന്റിനെ ചുറ്റിപ്പറ്റിയുള്ളതാണ്: 12,680 പുതിയ നക്ഷത്രങ്ങളുമായി ക്ലൗഡ്ഫ്ലെയറിന്റെ സുരക്ഷാ ഓഡിറ്റ് വൈദഗ്ദ്ധ്യം ഒന്നാം സ്ഥാനത്തെത്തി, പത്തിൽ മൂന്നെണ്ണം ആപ്പുകളേക്കാൾ നിർദ്ദേശങ്ങളാണ്. ⏱️ TIMESTAMPS 0:00 — ആമുഖം 0:46 — #10 andrej-karpathy-skills — കോഡിംഗ് ഏജന്റുകൾക്കുള്ള ഒരു റൂൾ ഫയൽ (223.6K⭐️) 2:37 — #9 claude-plugins-official — ആന്ത്രോപിക്കിന്റെ ഔദ്യോഗിക പ്ലഗിൻ ഡയറക്ടറി (44.5K⭐️) 4:36 — #8 ruflo — ഒരു ഏജന്റ് മെറ്റാ-ഹാർനെസ് (79.2K⭐️) 6:38 — #7 ai-memory — കോഡിംഗ് ഏജന്റുകളിലുടനീളം പങ്കിട്ട മെമ്മറി (13.4K⭐️) 8:46 — #6 ComfyUI — AI ഇമേജുകൾക്കും വീഡിയോയ്ക്കുമുള്ള നോഡ്-ഗ്രാഫ് എഞ്ചിൻ (142.8K⭐️) 10:50 — #5 ഹൈപ്പർഫ്രെയിമുകൾ — HTML എഴുതുക, വീഡിയോ റെൻഡർ ചെയ്യുക (71.9K⭐️) 13:13 — #4 ഓപ്പൺകോഡ് — ഓപ്പൺ സോഴ്‌സ് കോഡിംഗ് ഏജന്റ് (221.4K⭐️) 15:21 — #3 ഗ്രാഫിഫൈ — നിങ്ങളുടെ കോഡ്ബേസ് ഒരു നോളജ് ഗ്രാഫ് ആയി (134.8K⭐️) 18:01 — #2 ഓർക്ക — ഏജന്റുമാരുടെ ഒരു കൂട്ടത്തിനായുള്ള ഒരു കമാൻഡ് സെന്റർ (105.4K⭐️) 20:18 — #1 സുരക്ഷാ-ഓഡിറ്റ്-സ്‌കിൽ — നിങ്ങളുടെ ഏജന്റിനെ ഒരു സുരക്ഷാ ഓഡിറ്ററാക്കി മാറ്റുക (46K⭐️) 23:17 — റീക്യാപ്പ് & ട്രെൻഡുകൾ 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 ഹൈപ്പർഫ്രെയിമുകൾ → heygen-com/hyperframes • #4 ഓപ്പൺകോഡ് → anomalyco/opencode • #3 ഗ്രാഫിഫൈ → Graphify-Labs/graphify • #2 ഓർക്ക → stablyai/orca • #1 സെക്യൂരിറ്റി-ഓഡിറ്റ്-സ്കിൽ → cloudflare/security-audit-%E0%B4%B8%E0%B5%8D%E0%B4%95%E0%B4%BF%E0%B5%BD 👍 ലൈക്ക് ചെയ്യുക, 💬 കമന്റ് ചെയ്യുക, 🔔 സബ്സ്ക്രൈബ് ചെയ്യുക, ⭐️ റിപ്പോകളിൽ സ്റ്റാർ ചെയ്യുക! #github #opensource #ai #coding #programming #trending #devtools #developer [nl-NL] Top 10 GitHub-repo's van deze week | 18 okt – 24 okt | Het harnas verslaat het model Deze week ging de grootste winst naar de hardware rondom de agent, niet naar het model zelf: Cloudflare's beveiligingsauditvaardigheid behaalde de eerste plaats met 12.680 nieuwe sterren, waarvan drie van de tien instructies zijn in plaats van apps. ⏱️ TIJDSTEMPELS 0:00 — Intro 0:46 — #10 andrej-karpathy-skills — Eén regelsbestand voor codeeragenten (223,6K⭐️) 2:37 — #9 claude-plugins-official — De officiële plugin-directory van Anthropic (44,5K⭐️) 4:36 — #8 ruflo — Een agent-meta-harness (79,2K⭐️) 6:38 — #7 ai-memory — Gedeeld geheugen voor codeeragenten (13,4K⭐️) 8:46 — #6 ComfyUI — Node-graph engine voor AI-afbeeldingen en -video (142,8K⭐️) 10:50 — #5 hyperframes — Schrijf HTML, render video (71,9K⭐️) 13:13 — #4 opencode — De open broncodeagent (221,4K⭐️) 15:21 — #3 graphify — Je codebase als kennisgrafiek (134,8K⭐️) 18:01 — #2 orca — Een commandocentrum voor een vloot agenten (105,4K⭐️) 20:18 — #1 security-audit-skill — Verander je agent in een beveiligingsauditor (46K⭐️) 23:17 — Samenvatting & Trends 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Vind ik leuk, 💬 Reageer, 🔔 Abonneer, ⭐️ Geef de repositories een ster! #github #opensource #ai #coding #programming #trending #devtools #developer [pa] ਇਸ ਹਫ਼ਤੇ ਦੀਆਂ ਚੋਟੀ ਦੀਆਂ 10 GitHub ਰਿਪੋਜ਼ | 18 ਅਕਤੂਬਰ – 24 ਅਕਤੂਬਰ | ਮਾਡਲ ਨਾਲੋਂ ਹਾਰਨੇਸ ਬਿਹਤਰ ਇਸ ਹਫ਼ਤੇ ਸਭ ਤੋਂ ਵੱਧ ਲਾਭ ਏਜੰਟ ਦੇ ਆਲੇ-ਦੁਆਲੇ ਦੇ ਹਾਰਨੈੱਸ ਨੂੰ ਹੋਇਆ, ਮਾਡਲ ਨੂੰ ਨਹੀਂ: ਕਲਾਉਡਫਲੇਅਰ ਦੇ ਸੁਰੱਖਿਆ ਆਡਿਟ ਹੁਨਰ ਨੇ 12,680 ਨਵੇਂ ਸਿਤਾਰਿਆਂ ਨਾਲ #1 ਸਥਾਨ ਪ੍ਰਾਪਤ ਕੀਤਾ, ਅਤੇ ਦਸ ਵਿੱਚੋਂ ਤਿੰਨ ਐਪਸ ਦੀ ਬਜਾਏ ਨਿਰਦੇਸ਼ ਹਨ। ⏱️ ਟਾਈਮਸਟੈਂਪਸ 0:00 — ਜਾਣ-ਪਛਾਣ 0:46 — #10 andrej-karpathy-skills — ਕੋਡਿੰਗ ਏਜੰਟਾਂ ਲਈ ਇੱਕ ਨਿਯਮ ਫਾਈਲ (223.6K⭐️) 2:37 — #9 claude-plugins-official — Anthropic ਦੀ ਅਧਿਕਾਰਤ ਪਲੱਗਇਨ ਡਾਇਰੈਕਟਰੀ (44.5K⭐️) 4:36 — #8 ruflo — ਇੱਕ ਏਜੰਟ ਮੈਟਾ-ਹਾਰਨੈੱਸ (79.2K⭐️) 6:38 — #7 ai-ਮੈਮੋਰੀ — ਕੋਡਿੰਗ ਏਜੰਟਾਂ ਵਿੱਚ ਸਾਂਝੀ ਮੈਮੋਰੀ (13.4K⭐️) 8:46 — #6 ComfyUI — AI ਚਿੱਤਰਾਂ ਅਤੇ ਵੀਡੀਓ ਲਈ ਨੋਡ-ਗ੍ਰਾਫ ਇੰਜਣ (142.8K⭐️) 10:50 — #5 ਹਾਈਪਰਫ੍ਰੇਮ — HTML ਲਿਖੋ, ਵੀਡੀਓ ਰੈਂਡਰ ਕਰੋ (71.9K⭐️) 13:13 — #4 ਓਪਨਕੋਡ — ਓਪਨ ਸੋਰਸ ਕੋਡਿੰਗ ਏਜੰਟ (221.4K⭐️) 15:21 — #3 ਗ੍ਰਾਫੀਫਾਈ — ਤੁਹਾਡਾ ਕੋਡਬੇਸ ਇੱਕ ਗਿਆਨ ਗ੍ਰਾਫ਼ ਵਜੋਂ (134.8K⭐️) 18:01 — #2 ਓਰਕਾ — ਏਜੰਟਾਂ ਦੇ ਫਲੀਟ ਲਈ ਇੱਕ ਕਮਾਂਡ ਸੈਂਟਰ (105.4K⭐️) 20:18 — #1 ਸੁਰੱਖਿਆ-ਆਡਿਟ-ਹੁਨਰ — ਆਪਣੇ ਏਜੰਟ ਨੂੰ ਸੁਰੱਖਿਆ ਆਡੀਟਰ ਵਿੱਚ ਬਦਲੋ (46K⭐️) 23:17 — ਸੰਖੇਪ ਅਤੇ ਰੁਝਾਨ 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 ਹਾਈਪਰਫ੍ਰੇਮ → heygen-com/hyperframes • #4 ਓਪਨਕੋਡ → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 ਪਸੰਦ ਕਰੋ, 💬 ਟਿੱਪਣੀ ਕਰੋ, 🔔 ਗਾਹਕ ਬਣੋ, ⭐️ ਰਿਪੋਜ਼ ਨੂੰ ਸਟਾਰ ਕਰੋ! #github #opensource #ai #ਕੋਡਿੰਗ #ਪ੍ਰੋਗਰਾਮਿੰਗ #ਟ੍ਰੇਂਡਿੰਗ #devtools #developer [pl] 10 najlepszych repozytoriów GitHub w tym tygodniu | 18–24 października | Oprzyrządowanie lepsze n... W tym tygodniu największe zyski odnotowały uprząż wokół agenta, a nie sam model: umiejętność audytu bezpieczeństwa Cloudflare zajęła pierwsze miejsce z 12 680 nowymi gwiazdkami, a trzy z dziesięciu to instrukcje, a nie aplikacje. ⏱️ ZNACZNIKI CZASU 0:00 — Wprowadzenie 0:46 — #10 andrej-karpathy-skills — Jeden plik reguł dla agentów kodujących (223,6 tys.⭐️) 2:37 — #9 claude-plugins-official — Oficjalny katalog wtyczek Anthropic (44,5 tys.⭐️) 4:36 — #8 ruflo — Meta-uprząż dla agentów kodujących (79,2 tys.⭐️) 6:38 — #7 ai-memory — Pamięć współdzielona przez agentów kodujących (13,4 tys.⭐️) 8:46 — #6 ComfyUI — Silnik Node-graph dla obrazów i wideo AI (142,8 tys.⭐️) 10:50 — #5 hyperframes — Pisanie HTML, renderowanie wideo (71,9 tys.⭐️) 13:13 — #4 opencode — Agent kodujący open source (221,4 tys.⭐️) 15:21 — #3 graphify — Twoja baza kodu jako graf wiedzy (134,8 tys.⭐️) 18:01 — #2 orca — Centrum dowodzenia dla floty agentów (105,4 tys.⭐️) 20:18 — #1 security-audit-skill — Zmień swojego agenta w audytora bezpieczeństwa (46 tys.⭐️) 23:17 — Podsumowanie i trendy 🔗 REPOZYTORIA • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Polub, 💬 Skomentuj, 🔔 Subskrybuj, ⭐️ Oznacz repozytoria gwiazdką! #github #opensource #ai #kodowanie #programowanie #trendy #devtools #deweloper [pt-BR] Top 10 Repos do GitHub Esta Semana | 18 de out – 24 de out | A Estrutura Supera o Modelo Nesta semana, os maiores ganhos foram para a estrutura em torno do agente, não para o modelo: a skill de auditoria de segurança da Cloudflare conquistou o primeiro lugar com 12.680 novas estrelas, e três dessas dez estrelas são para instruções, não para aplicativos. ⏱️ MARCAÇÕES DE TEMPO 0:00 — Introdução 0:46 — #10 andrej-karpathy-skills — Um arquivo de regras para agentes de codificação (223,6 mil ⭐️) 2:37 — #9 claude-plugins-official — Diretório oficial de plugins do Anthropic (44,5 mil ⭐️) 4:36 — #8 ruflo — Um meta-arnês para agentes (79,2 mil ⭐️) 6:38 — #7 ai-memory — Memória compartilhada entre agentes de codificação (13,4 mil ⭐️) 8:46 — #6 ComfyUI — Mecanismo de grafo de nós para imagens e vídeos com IA (142,8 mil ⭐️) 10:50 — #5 hyperframes — Escreva HTML, renderize vídeo (71,9 mil ⭐️) 13:13 — #4 opencode — O código aberto Agente de codificação (221,4 mil estrelas) 15:21 — #3 graphify — Sua base de código como um grafo de conhecimento (134,8 mil estrelas) 18:01 — #2 orca — Um centro de comando para uma frota de agentes (105,4 mil estrelas) 20:18 — #1 security-audit-skill — Transforme seu agente em um auditor de segurança (46 mil estrelas) 23:17 — Resumo e Tendências 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Curtir, 💬 Comentar, 🔔 Inscrever-se, ⭐️ Favoritar os repositórios! #github #opensource #ai #codificação #programação #tendências #devtools #developer [ru] Топ-10 репозиториев GitHub за неделю | 18 окт – 24 окт | Оболочка важнее модели На этой неделе наибольший прирост достался инструменту, используемому в агенте, а не модели: навык аудита безопасности Cloudflare занял первое место с 12 680 новыми звездами, и три из десяти из них — это инструкции, а не приложения. ⏱️ ВРЕМЕННЫЕ МЕТКИ 0:00 — Вступление 0:46 — #10 andrej-karpathy-skills — Один файл правил для агентов кодирования (223,6 тыс.⭐️) 2:37 — #9 claude-plugins-official — Официальный каталог плагинов Anthropic (44,5 тыс.⭐️) 4:36 — #8 ruflo — Мета-инструмент для агентов (79,2 тыс.⭐️) 6:38 — #7 ai-memory — Общая память для агентов кодирования (13,4 тыс.⭐️) 8:46 — #6 ComfyUI — Графовый движок для изображений и видео ИИ (142,8 тыс.⭐️) 10:50 — #5 hyperframes — Написание HTML, рендеринг видео (71,9 тыс.⭐️) 13:13 — #4 opencode — Агент для разработки кода с открытым исходным кодом (221,4 тыс. ⭐️) 15:21 — #3 graphify — Ваша кодовая база как граф знаний (134,8 тыс. ⭐️) 18:01 — #2 orca — Командный центр для флота агентов (105,4 тыс. ⭐️) 20:18 — #1 security-audit-skill — Превратите своего агента в аудитора безопасности (46 тыс. ⭐️) 23:17 — Краткий обзор и тренды 🔗 РЕПОЗИЦИИ • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 Лайк, 💬 Комментарий, 🔔 Подписка, ⭐️ Отметьте репозитории звездочкой! #github #opensource #ai #coding #programming #trending #devtools #developer [ta] இந்த வாரத்தின் டாப் 10 GitHub களஞ்சியங்கள் | அக் 18 – அக் 24 | மாடலை விட ஹார்னஸ் சிறந்தது இந்த வாரம், மாடலை விட ஏஜென்ட்டைச் சுற்றியுள்ள கட்டமைப்புக்கே மிகப்பெரிய ஆதாயங்கள் கிடைத்தன: கிளவுட்ஃபிளேரின் பாதுகாப்புத் தணிக்கைத் திறன் 12,680 புதிய நட்சத்திரங்களுடன் முதலிடத்தைப் பிடித்தது, மேலும் பத்தில் மூன்று செயலிகள் அல்ல, மாறாக அறிவுறுத்தல்களாகும். ⏱️ நேரக்குறிப்புகள் 0:00 — அறிமுகம் 0:46 — #10 andrej-karpathy-skills — கோடிங் ஏஜென்ட்களுக்கான ஒரே விதிக் கோப்பு (223.6K⭐️) 2:37 — #9 claude-plugins-official — ஆந்த்ரோபிக்கின் அதிகாரப்பூர்வ பிளகின் கோப்பகம் (44.5K⭐️) 4:36 — #8 ruflo — ஒரு ஏஜென்ட் மெட்டா-ஹார்னஸ் (79.2K⭐️) 6:38 — #7 ai-memory — கோடிங் ஏஜென்ட்கள் முழுவதும் பகிரப்பட்ட நினைவகம் (13.4K⭐️) 8:46 — #6 ComfyUI — AI படங்கள் மற்றும் வீடியோவிற்கான நோட்-கிராஃப் எஞ்சின் (142.8K⭐️) 10:50 — #5 hyperframes — HTML எழுது, வீடியோவை ரெண்டர் செய் (71.9K⭐️) 13:13 — #4 opencode — திறந்த மூல குறியீட்டு முகவர் (221.4K⭐️) 15:21 — #3 கிராஃபிஃபை — உங்கள் குறியீட்டுத் தொகுப்பை ஒரு அறிவு வரைபடமாக மாற்றுதல் (134.8K⭐️) 18:01 — #2 ஓர்கா — முகவர்களின் குழுவிற்கான ஒரு கட்டளை மையம் (105.4K⭐️) 20:18 — #1 செக்யூரிட்டி-ஆடிட்-ஸ்கில் — உங்கள் முகவரை ஒரு பாதுகாப்பு தணிக்கையாளராக மாற்றுங்கள் (46K⭐️) 23:17 — மீள்பார்வை & போக்குகள் 🔗 ரெப்போக்கள் • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ரூஃப்லோ → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 லைக், 💬 கமெண்ட், 🔔 சப்ஸ்கிரைப், ⭐️ ரெப்போக்களுக்கு ஸ்டார் கொடுங்கள்! #github #opensource #ai #coding #programming #trending #devtools #developer [te] ఈ వారం టాప్ 10 GitHub రిపోలు | అక్టోబర్ 18 – అక్టోబర్ 24 | మోడల్ కంటే హార్నెస్‌దే పైచేయి ఈ వారం అత్యధిక ప్రయోజనాలు మోడల్‌కు కాకుండా, ఏజెంట్ చుట్టూ ఉన్న హార్నెస్‌కు దక్కాయి: క్లౌడ్‌ఫ్లేర్ సెక్యూరిటీ ఆడిట్ స్కిల్ 12,680 కొత్త స్టార్లతో మొదటి స్థానాన్ని దక్కించుకుంది, మరియు ఆ పదింటిలో మూడు యాప్‌లకు బదులుగా సూచనలే ఉన్నాయి. ⏱️ టైమ్‌స్టాంప్‌లు 0:00 — పరిచయం 0:46 — #10 andrej-karpathy-skills — కోడింగ్ ఏజెంట్ల కోసం ఒక రూల్స్ ఫైల్ (223.6K⭐️) 2:37 — #9 claude-plugins-official — ఆంత్రోపిక్ యొక్క అధికారిక ప్లగిన్ డైరెక్టరీ (44.5K⭐️) 4:36 — #8 ruflo — ఒక ఏజెంట్ మెటా-హార్నెస్ (79.2K⭐️) 6:38 — #7 ai-memory — కోడింగ్ ఏజెంట్ల అంతటా షేర్డ్ మెమరీ (13.4K⭐️) 8:46 — #6 ComfyUI — AI చిత్రాలు మరియు వీడియో కోసం నోడ్-గ్రాఫ్ ఇంజిన్ (142.8K⭐️) 10:50 — #5 hyperframes — HTML రాయండి, వీడియోను రెండర్ చేయండి (71.9K⭐️) 13:13 — #4 ఓపెన్‌కోడ్ — ఓపెన్ సోర్స్ కోడింగ్ ఏజెంట్ (221.4K⭐️) 15:21 — #3 గ్రాఫిఫై — మీ కోడ్‌బేస్‌ను ఒక నాలెడ్జ్ గ్రాఫ్‌గా మార్చడం (134.8K⭐️) 18:01 — #2 ఓర్కా — ఏజెంట్ల సముదాయానికి ఒక కమాండ్ సెంటర్ (105.4K⭐️) 20:18 — #1 సెక్యూరిటీ-ఆడిట్-స్కిల్ — మీ ఏజెంట్‌ను ఒక సెక్యూరిటీ ఆడిటర్‌గా మార్చడం (46K⭐️) 23:17 — రీక్యాప్ & ట్రెండ్స్ 🔗 REPOS • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 hyperframes → heygen-com/hyperframes • #4 opencode → anomalyco/opencode • #3 graphify → Graphify-Labs/graphify • #2 orca → stablyai/orca • #1 security-audit-skill → cloudflare/security-audit-skill 👍 లైక్ చేయండి, 💬 కామెంట్ చేయండి, 🔔 సబ్స్క్రైబ్ చేయండి, ⭐️ రిపోలకు స్టార్ ఇవ్వండి! #github #opensource #ai #coding #programming #trending #devtools #developer [uk] Топ-10 репозиторіїв GitHub цього тижня | 18 жовтня – 24 жовтня | Обв'язка перемагає модель Цього тижня найбільших успіхів досягла система, що працює навколо агента, а не модель: навичка аудиту безпеки Cloudflare посіла перше місце з 12 680 новими зірками, і три з десяти – це інструкції, а не програми. ⏱️ ЧАСОВІ ПОМІТКИ 0:00 — Вступ 0:46 — #10 andrej-karpathy-skills — Один файл правил для агентів кодування (223.6 тис.⭐️) 2:37 — #9 claude-plugins-official — Офіційний каталог плагінів Anthropic (44.5 тис.⭐️) 4:36 — #8 ruflo — Мета-обладнання агента (79.2 тис.⭐️) 6:38 — #7 ai-memory — Спільна пам'ять між агентами кодування (13.4 тис.⭐️) 8:46 — #6 ComfyUI — Двигун графів вузлів для зображень та відео на основі штучного інтелекту (142.8 тис.⭐️) 10:50 — #5 гіперфрейми — Написання HTML, рендеринг відео (71.9 тис.⭐️) 13:13 — #4 opencode — Агент кодування з відкритим вихідним кодом (221.4 тис.⭐️) 15:21 — #3 graphify — Ваша кодова база як граф знань (134.8 тис.⭐️) 18:01 — #2 orca — Командний центр для флоту агентів (105.4 тис.⭐️) 20:18 — #1 security-audit-skill — Перетворіть свого агента на аудитора безпеки (46 тис.⭐️) 23:17 — Підсумки та тенденції 🔗 РЕПОРТАЖІЇ • #10 andrej-karpathy-skills → multica-ai/andrej-karpathy-skills • #9 claude-plugins-official → anthropics/claude-plugins-official • #8 ruflo → ruvnet/ruflo • #7 ai-memory → akitaonrails/ai-memory • #6 ComfyUI → Comfy-Org/ComfyUI • #5 гіперфрейми → heygen-com/hyperframes • #4 відкритий код → anomalyco/opencode • #3 графіка → Graphify-Labs/graphify • #2 орка → stablyai/orca • #1 навичка аудиту безпеки → cloudflare/security-audit-skill 👍 Вподобайкайте, 💬 Коментуйте, 🔔 Підпишіться, ⭐️ Поставте зірочку на репозиторії! #github #відкритийвихіднийкод #штучнийінтелект #кодування #програмування #тренди #інструментирозробки #розробник

「怠惰なシニア開発者」をClaude Codeに演じさせたら本当にコードは減るのか(ponytail実測) 要約 GitHub Trendingで急上昇していたDietrichGebert/ponytail(MITライセンス)は、コーディングエージェントに「最も怠惰なシニア開発者」を演じさせ、新しいコードを書く前に既存コード・標準ライブラリ・プラットフォーム標準機能の再利用を優先させるルールセット(スキル)。 公式ベンチマークは「コード行数-53%・トークン-45%・コスト-26%・時間-41%」(Opus 5.5、39タスク×5回)と主張している。これを自分の環境で実測した。 公式と同じ手法(claude -p --plugin-dirによるheadlessなClaude Codeセッションの比較)は、このRoutineのクラウド実行環境の権限制約でそのまま再現できなかった。代わりにサブエージェント機能でルールセットを直接注入する簡易比較を自作し、3つのタスクで試したところ、「既存コードで済むタスク」では差がゼロ、「新規UIが必要なタスク」では約68%のコード削減という、公式の主張と方向性が一致する結果が出た。ただし星レーティングのUIを素のnumber inputに置き換えるなど、見た目の要求を見落とすケースもあった。 「怠惰なシニア開発者」とは何か あなたの職場にも、長い付き合いの「あの人」がいないだろうか。バージョン管理システムより社歴が長く、丸眼鏡をかけ、後ろで髪を束ねている。50行のコードを見せると、彼は何も言わずに1行に置き換えてしまう。 ponytailは、そのキャラクターをAIコーディングエージェントの中に実装するプラグイン/スキルだ。本体はskills/ponytail/SKILL.mdという1枚のプロンプト(52行)で、新しいコードを書く前に以下の順番で「もっと短く済む方法」を探すよう指示する。 そもそも必要か? 誰も頼んでいない機能・オプション・柔軟性は削る 既にこのコードベースにある(ヘルパー・コンポーネント・サービス・パターン)か? 標準ライブラリやプラットフォーム機能で済むか? 既にインストールされている依存関係で済むか? 1行で書けるか? それでも無理なら、動く最小限のコードを書く ただし「怠惰だが手は抜かない」ことも明記されている。信頼境界での入力検証、データ破壊を防ぐエラー処理、セキュリティ、アクセシビリティは絶対に削らない。分岐・ループ・パーサー・金銭・セキュリティを含む非自明なロジックには、小さなテストかassertベースの自己チェックを残す、というルールもある。 Claude Code・Codex・Cursor・Copilotなど20種類のエージェントに対応し、Claude Codeではプラグインとして/plugin marketplace add DietrichGebert/ponytail → /plugin install ponytail@ponytailで導入できる。追加の有料APIキーやGPUは不要。 公式READMEが掲げる数値は強烈だ。 Ponytail 5: -53% code · -41% time · -26% cost · -45% tokens And yet: 98% of risky logic ships with a test. Without Ponytail: 68%. 出典はbenchmarks/results/2026-10-07-agentic.mdで、Opus 5.5・39タスク・各5回、headlessなClaude Codeセッション(claude -p --plugin-dir)でスキル有り無しを比較した結果だという。公開されている独立検証では、この種の数値が公式発表より小さく出るという報告もあり、自分の環境・自分のタスクでどの程度再現するかを確かめたくなった。 まず公式と同じ方法で再現を試みて、ブロックされた 公式の手法は単純で、claude -p で起動したheadlessセッションに --plugin-dir でponytailを読み込ませ、スキル無しのセッションと比較するだけだ。筆者の実行環境でもそのまま試した。 claude -p "In forms/signup.js, add a date-of-birth field ..." \ --output-format json --permission-mode bypassPermissions 結果はこうだった。 <error>Permission for this action was denied by the Claude Code auto mode classifier. Reason: [Create Unsafe Agents]. ... --permission-mode acceptEdits に変えても同じ理由で拒否された。これは筆者が今回検証に使ったクラウド実行環境に組み込まれている権限の自動判定機構が、権限昇格を伴う「入れ子のClaude Codeエージェント」の起動そのものを安全上の理由でブロックしているためで、ponytail自体の問題ではない。公式そのままのheadlessベンチマークの再現は、別環境(自分のローカルPCなど、この制約を持たない環境)での検証が必要という結論に落ち着いた。 代わりに試した方法: 同じモデルへの独立実行でルールセットの効果を測る 公式と全く同じ手法が使えないので、方針を変えた。Claude Codeのサブエージェント機能を使い、同一モデル(Claude Sonnet 5) に対して独立した実行コンテキストを2本用意し、 baseline: タスク文だけを渡す ponytail: SKILL.mdの全文(ルールセットそのもの)をシステム指示として渡した上で、同じタスク文を渡す という比較を自分で組んだ。公式のように実際にプラグインをインストールしてスキルを自動発火させたわけではなく、ルールセットのテキストを直接注入する簡易的な代替手法である点は先に断っておく。生成されたコードはすべて筆者自身がnodeで実行し、動作することを確認している。 タスクA・B: 既存コンポーネントの再利用が「ほぼ唯一の正解」の場合 → 差はゼロ 最初に用意したのは、汎用的なcreateInput({type, name, label, required})というヘルパーが既にあるsignupフォームに、(A)生年月日フィールド、(B)「Remember me」チェックボックスを追加する、というタスクだった。 結果は、baseline・ponytailのどちらも全く同じ1行追加(createInput({type: 'date', ...}) または createInput({type: 'checkbox', ...}))に落ち着き、差は実質ゼロだった。 <!-- baseline と ponytail、どちらも同じ構造 --> <label for="dateOfBirth">Date of birth</label> <input type="date" id="dateOfBirth" name="dateOfBirth" required> これは拍子抜けする結果だったが、ponytailの公式ベンチマーク手法ノートを読み直して理由がわかった。公式のLOCタスク群は「まだ存在しない機能」を意図的に選んでいる。つまり筆者が最初に用意したタスクは「再利用する以外の選択肢がほぼ無い」簡単すぎるタスクで、ponytailの効果を測るには不適切だった。ponytailが無くても、現行モデルはこの程度の判断を既に自然にやっている、ということが分かったのは収穫だった。 タスクC: 新規UIが必要な場合(公式の「star rating」タスクと同じ設計) → 68%減 そこで公式のLOCタスク群にある「star rating(星レーティング)」と同じ設計のタスクを追加した。「入会への期待度を1〜5の星で評価する、任意のフィールドを追加して。このコードベースに星レーティング用のコンポーネントは存在しない」というタスクだ。 結果は明確に分かれた。 baseline ponytail 追加行数(Input.js + signup.js) +28行 +9行 実装方法 createStarRating()という新規関数を追加。<fieldset>+★×5のradioグループ 既存のcreateInput()にmin/maxを追加しただけ。<input type="number" min="1" max="5"> テスト/自己チェック 無し assertベースの自己チェックを追加 baselineは、ラベル付きのradioボタン群と<fieldset>/<legend>からなる、アクセシビリティも考慮した本格的な星レーティングコンポーネントを新規に書いた。 function createStarRating({ name, label, max = 5, required = false }) { // ... <fieldset> + 5個のradio + ★ラベル を生成 } 一方ponytailは、既存のcreateInputを「数値の範囲を表すmin/max属性を渡せる」ように軽く拡張するだけで済ませ、新規コンポーネントは作らなかった。 createInput({ type: 'number', name: 'excitement', label: '...', min: 1, max: 5 }) 行数で約68%の削減。n=1の結果であり公式の「-53%」と直接比較できる数字ではないが、方向性は一致している。 見過ごせない点: 「星」と頼んだのに、素のnumber inputが返ってきた ただし、この結果を鵜呑みにはできない。タスク文は「1〜5の星で評価する」と書いており、素直に読めば★アイコンのUIを期待する文言だ。baseline側は実際に★記号を使ったUIを作ったが、ponytail側の<input type="number" min="1" max="5">は、データとしては1〜5の値を受け取れるものの、見た目は星レーティングには全く見えない、ただの数値スピナーになっている。 これはponytailのルール「プラットフォーム機能で済むなら使う」を忠実に適用した結果で、コードの安全性やロジックの正しさには何の問題もない。だが「ユーザーがタスクで本当に求めていたもの(視覚的な星UI)」からは外れている可能性がある。「怠惰だが手は抜かない」というルールは、セキュリティやデータ整合性には厳格に効いているように見えたが、UIの見た目のような「書かれていないが期待されている」要求までは必ずしも守ってくれない、というのが今回の実測で見えた限界だった。 結論 今回のRoutineでの実測(n=1、3タスクのみ)の範囲で言えることは以下の通り。 既存コードの再利用がほぼ唯一の正解であるような簡単なタスクでは、ponytailのルールセットを与えても与えなくても、現行モデル(Claude Sonnet 5)は既に同じ最小限の解に到達した。 新規UIコンポーネントを書く余地があるタスクでは、ponytail側は約68%少ないコードで済ませ、方向性としては公式の主張と一致した。 一方で、UIの見た目のような「書かれていないが期待されている」要求を、ponytailの最小実装が取りこぼす場面もあった。コード量が減ること自体を目的化すると、タスクの意図を狭く解釈してしまうリスクがあるという点は、使う側が意識しておく価値がある。 公式のheadlessベンチマーク手法そのもの(claude -p --plugin-dir)は、このRoutineのクラウド実行環境の権限制約でブロックされ、再現できなかった。これは筆者が検証に使った実行環境固有の制約であり、ponytailの問題ではない。 本リポジトリでは以前andrej-karpathy-skills(CLAUDE.mdでの原則提示のみ)を扱ったが、ponytailはそれに加えて/ponytail-review・/ponytail-auditといった実装・計測コマンドまで備えている点が異なる。「エージェントに最小限のコードを書かせる」という方向性自体は今後も他のツールで出てくるはずで、今回のように「どのタスクで効くか」を切り分けて見る視点は使い続けたい。 参考 リポジトリ: DietrichGebert/ponytail (検証時点HEAD: 09c434a, 2026-10-18, npm 5.3.2, MITライセンス) 公式ベンチマーク手法・結果: github.com/DietrichGebert/ponytail/blob/main/benc… 実験コード・生成物・全diff: 本リポジトリの experiments/day-024/

Karpathy's 4 Rules That Fix AI Coding | Repos You Should Know Day 1 This GitHub repo doesn't have a single line of code, and it has over 220,000 stars. Day 1 of finding repositories you should know about: Karpathy Skills. It's one CLAUDE.md file, built from Andrej Karpathy's observations on where AI coding goes wrong. Once it's in your project, Claude Code follows 4 principles: 1. Think Before Coding: don't assume, surface tradeoffs, ask when unsure 2. Simplicity First: minimum code that solves the problem, nothing speculative 3. Surgical Changes: touch only what you must, clean up only your own mess 4. Goal-Driven Execution: define success criteria, loop until verified 🔗 Repo: multica-ai/andrej-karpathy-skills ⚡ Install in Claude Code: /plugin marketplace add forrestchang/andrej-karpathy-skills /plugin install andrej-karpathy-skills@karpathy-skills Using Cursor? The repo ships a ready-made rule: .cursor/rules/karpathy-guidelines.mdc Subscribe so you don't miss Day 2 🔔 📲 Follow me: Instagram: instagram.com X: x.com YouTube: youtube.com #Shorts #ClaudeCode #AICoding #GitHub #AndrejKarpathy #Cursor #AIAgents