Software engineering
471 tweets
This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over. Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it.
NEWS: Boston Dynamics has just released a new video of its upgraded next-generation humanoid robot called Atlas. • 4 hour battery. Self-swappable for continuous operation • 6 feet 2 inches tall • Weight: 198 lbs • 56 total degrees of freedom • Now fully electric, ditching older hydraulic systems • New lightweight mix of aluminum and titanium components • 110 lbs weight capacity (66 lbs sustained) • Can reach up to 7.5 ft • Constantly evaluates its surroundings and adjusts its posture, balance, and grip in real time • Hands that can reconfigure as needed. Tactile sensors feed data back into the system, helping apply the right amount of force • Brain is powered by Nvidia chips
We’ve merged Slash Commands into Skills in Claude Code. You do not need to do anything to migrate to this and it should not disrupt any of your existing workflows. You can invoke any skill with the slash command syntax by starting with `/`. Similarly, every slash command you currently have can be called as a Skill by Claude Code. Additionally, you can use subagents with Skills seamlessly. ## Why Combine Slash Commands and Skills? Slash Commands were one of our first abstractions for managing context, and served as a form of progressive disclosure. You could make sure the model only loaded this context when needed. But as model capabilities have advanced, we realized Skills were the more powerful way of loading context. They allow the model to load in context dynamically by reading relevant files and you could reference other files inside of your SKILL.MD which would allow for multiple levels of dynamic context. We call this progressive disclosure and you can read more about how this works with Skills in our engineering blog. Combining skills & slash makes it easier for Claude (e.g. it doesn’t need a SlashCommand Tool and a Skill Tool) and it also simplifies the user mental model. However, there is no migration needed. Your slash commands in ~/.claude/commands will continue to work as normal. Going forward, when thinking of making a slash command we suggest making a skill instead. This will allow you to use new extensions we add to Skills, such as the ones to work with subagents. For each skill you create, you can choose whether you want it to be invocable, model-invocable, or both (the default). If you do not want a user to be able to invoke a skill with a slash command, you can set user-invocable: false. If you do not want the model to invoke a slash command automatically you can set disable-model-invocation: true # Using subagents with Skills Skills naturally pair with subagents. Subagents allow you to execute the skill while protecting your context window, you can also choose which subagent is activated and if you want to fork the context. Here are some examples of when you might use that: ## Search Skills with the Explore Agent Setting agent: <agent-name> will spawn a subagent that loads the skill into its context. Search is a great example of a type of skill where you might want to use a subagent. For example, a Research skill where you want to summarize a set of files using the Explore agent and return it. ## Memory Skills With Forked Context Setting context: fork spins off a subagent which has all of your current context. This is great when you want to do an operation in parallel to your current work. For example, you might have a memory skill to summarize your most recent conversation and put it in a specific file. You wouldn't want these tool results to be in your main context though, because it’s not related to the work being done. Read all about this and more on Skills in our documentation: https://code.claude.com/docs/en/skills
Claude Code 2.1.0 is officially out! claude update to get it We shipped: - Shift+enter for newlines, w/ zero setup - Add hooks directly to agents & skills frontmatter - Skills: forked context, hot reload, custom agent support, invoke with / - Agents no longer stop when you deny a tool use - Configure the model to respond in your language (eg. Japanese, Spanish) - Wildcard support for tool permissions: eg. Bash(*-h*) - /teleport your session to http://claude.ai/code - Overall: 1096 commits https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md… If you haven't tried Claude Code yet: https://code.claude.com/docs/en/setup Lmk what you think!
The more I have AI agents write all my code, the more I feel that us devs will be alright (and possibly more in-demand for important stuff) Hard for me to imagine anyone building *reliable* software without an understanding of how to do this (either via experience or study)
Introducing Context7 Skills! ◆ We extracted 24k skills from 65k repos ◆ Skills for Tailwind, React, Better-Auth, etc. ◆ Install in a single CLI command Perfect for Cursor, Claude Code & others
Supabase now has Agent Skills! Install it on your favorite AI agent/ editor to have it follow the best practices for handling Postgres! $ npx skills add supabase/agent-skills
Introducing the first vibe coding platform that builds agentic AI apps. Build AI that thinks, searches the web, executes code, and completes tasks end-to-end. One-shots Cursor, Perplexity, Gamma, and more. In celebration of our launch, we're giving away a free month to everyone who replies. Repost and reply “Blink” below and we'll DM you the credits
Today, we're upgrading Todos in Claude Code to Tasks. Tasks are a new primitive that help Claude Code track and complete more complicated projects and collaborate on them across multiple sessions or subagents. As model capabilities grow, one of the most important things we can do is “unhobble” Claude and allow it to use its new capabilities effectively. Compared to previous models, Opus 4.5 is able to run autonomously for longer and keep track of its state better. We found that the TodoWrite Tool was no longer necessary because Claude already knew what it needed to do for smaller tasks. At the same time, we found ourselves using Claude Code to complete longer projects, sometimes across multiple subagents, context windows or sessions. But projects are more complex, tasks have dependencies and blockers and require coordination when using it across sessions. It was clear we needed to evolve Todos to help Claude work on longer projects. This need was also emerging in the community and we took inspiration from projects like Beads by Steve Yegge. Tasks are our new abstraction for coordinating many pieces of work across projects, Claude can create Tasks with dependencies on each other that are stored in the metadata, which mirrors more how projects work. Additionally, Tasks are stored in the file system so that multiple subagents or sessions can collaborate on them. When one session updates a Task, that is broadcasted to all sessions currently working on the same Task List. You can ask Claude to create tasks right now, it’s especially useful when creating when spinning up subagents. Tasks are stored in ~/.claude/tasks, you can use this to build additional utilities on top of tasks as well. To make sessions collaborate on a single Task List, you can set the TaskList as an environment variable and start Claude like so: CLAUDE_CODE_TASK_LIST_ID=groceries claude This also works for claude -p and the AgentSDK. Tasks are a key building block for allowing Claude to build more complex projects. We’re looking forward to seeing how you use it.
TLDR: The magic of Compound Engineering in @claudeai Code... building a X/Twitter (interface) clone in 20 mins. A massive game changer for me. (watch at 2x)
Had Claude Code build a little plugin that visualizes the work Claude Code is doing as agents working in an office, with agents doing work and passing information to each other. New subagents are hired, they acquire skills, and they turn in completed work. Fun start.
Damn, if you want your agent to do browser testing, you have to try agent-browser. The new version with it's updated SKILL is so fast and token efficient. Wow. Tell your agent this: Install/update agent-browser and update my skill file: https://github.com/vercel-labs/agent-browser/… https://github.com/vercel-labs/agent-browser/blob/main/skills/agent-browser/SKILL.md…
GitHub - vercel-labs/agent-browser: Browser automation CLI for AI agents
They call it getting “Claude-pilled.” It’s the moment software engineers, executives and investors turn their work over to Anthropic’s Claude AI—and then witness a thinking machine of shocking capability, even in an age awash in powerful AI tools. https://on.wsj.com/4r0nAwl
We will make the new 𝕏 algorithm, including all code used to determine what organic and advertising posts are recommended to users, open source in 7 days. This will be repeated every 4 weeks, with comprehensive developer notes, to help you understand what changed.
Introducing json-render AI-generated UI. Deterministic output. 1. Define your component catalog 2. AI steams JSON 3. Render interactive UI Let users prompt dashboards, widgets and apps - safely constrained to components and actions you define
Agent Skills are now available in Google Antigravity! Skills are an open standard to extend what your agent can do. Whether it's project-specific workflows or global utilities, you can now package knowledge into reusable skills.
I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.