In partnership with
Opening
I opened my terminal this morning and found herdr sitting at 13,414 stars. An agent multiplexer that lives in your terminal and routes work across multiple agents from one interface. No fanfare. No blog post. Just GitHub doing what it does: the community votes with stars, and something real rises.
That's the pattern I keep noticing. The tools that actually change how you build don't arrive with a press release. They arrive on a Tuesday with a commit and a README.
This issue: 10 drops including the agent multiplexer the community is crowding around, a repo that gives Claude eyes for video, five skills built specifically for Claude Code, a Stack pick that debugs your web apps from inside your agent, and three Signals worth your attention. Anthropic shipped Claude Cowork to mobile. Simon Willison dropped sqlite-utils 4.0 with real schema migrations. And Microsoft is betting its own models are good enough now.
Lead with the artifact, always.
advertisment
7 Proven External Traffic Strategies
Most eCommerce brands running external traffic aren't scaling — they're just spending.
Wrong channels, no real attribution, and at the end of the month, still no clear answer to the question that matters: what actually drove revenue?
The brands getting it right aren't necessarily spending more. They've just stopped guessing. They know which channels pull weight on Amazon listings, which ones bleed budget, and why affiliate and creator traffic outperforms on ROI when it's set up correctly.
Levanta put together a free playbook breaking down 7 proven external traffic strategies — where each one works, where it falls apart, and what it takes to scale without it becoming a second job.
Inside you'll see how top brands are driving millions in off-Amazon revenue and why most channels underdeliver when brands don't know what to look for.
If you're serious about growing outside of PPC, this is worth 5 minutes.
The Drops
[Repo] herdr, An agent multiplexer that lives in your terminal. Routes work across multiple agents from a single interface, 13,414 stars and trending now. If you're running more than two agents and context-switching manually, this is the fix.
[Repo] RuView, Turns commodity WiFi signals into real-time spatial intelligence: presence detection, vital sign monitoring, all without a camera pixel. 78,390 stars. The use case for operators building ambient-aware agents or occupancy tools is immediate.
[Repo] bradautomates/claude-video, Gives Claude the ability to watch any video. /watch downloads it, extracts frames, transcribes audio, and hands the whole thing to Claude for analysis. 5,040 stars. The gotcha: frame extraction gets expensive fast on long videos; set a frame-rate cap before you run it on anything over ten minutes.
[Repo] langgraph, Graph-based agent orchestration from LangChain: stateful multi-step loops, conditional branching, parallel node execution. The standard choice when your agent needs to hold state across more than two tool calls.
[Repo] aider, Terminal AI pair-programmer, git-aware, multi-file edits, architect mode. You feed it a task, it writes the diff and commits. The architect mode is the part most people miss: it plans the change before writing a line.
[Skill] udecode/dotai, A skills-and-agents framework for Claude Code, 1,151 stars. Gives you a structured way to build and chain reusable agent capabilities without rebuilding the plumbing each time.
[Skill] ZeroZ-lab/cc-design, High-fidelity HTML design and prototype guidance skill for AI agents, 812 stars. Closes the gap between "Claude writes some HTML" and "Claude ships something that looks intentional."
[Skill] hellowind777/hello2cc, Native-first Claude Code plugin that injects third-party models silently into your agent's model routing, 704 stars. Useful when you want Claude Code's interface but a cheaper model doing the grunt work.
[Skill] chrisbanes/skills, Skills for Kotlin, Jetpack Compose, and Android development, 823 stars. If you build Android in Claude Code and you don't have this, you're writing boilerplate Claude already solved.
[Skill] op7418/Youtube-clipper-skill, A Claude Code skill for clipping and extracting from YouTube videos, 2,045 stars. Pairs directly with the claude-video repo above if you want a full video-analysis workflow inside one session.
advertisment
10 AI Stocks to Lead the Next Decade
AI isn’t a tech trend – it’s a full-blown, multi-trillion dollar race, and 10 companies are already pulling ahead.
These are the innovators driving real revenue, attracting institutional attention, and positioning for massive growth.
Get all 10 tickers in The 10 Best AI Stocks to Own in 2026, free today.
The Stack
[MCP] kogiqa-mcp, A web browser MCP designed specifically for debugging and developing complex web apps from inside your agent. Not a generic browser tool: it's built to surface the stack traces and DOM state your agent needs to actually fix something, not just describe it. Install it, connect it to a project, and stop pasting error messages by hand.
advertisment
Hiring in a new country? Read this first.
Every country has different employment laws, payroll requirements, notice periods, and benefits expectations.
Oyster's Global Hiring Guides help you navigate the details, avoid surprises, and hire with confidence—wherever you're growing next.
Today's Signals
- Claude Cowork hits mobile and web. Anthropic launched Claude Cowork on mobile and web for Max subscribers first, broader rollout to follow. The operator consequence: long-running tasks now persist after you close your laptop, and the notification lands on your phone. Async agent work just got a new surface. (The Verge)
- sqlite-utils 4.0 ships with schema migrations. Simon Willison's 124th release of the project and its first major version bump since 2020. The addition that matters for operators: database schema migrations, meaning you can evolve a live SQLite database without a teardown-rebuild cycle. If you're using SQLite as a lightweight agent memory store or pipeline scratch pad, upgrade. (Simon Willison)
- Microsoft is switching to its own models. Microsoft is cutting back on external AI spend and routing more workloads to its in-house models. The directional signal for operators: inference costs are becoming a line item that even the biggest spenders are renegotiating. Every lab's pricing use shrinks a little when their largest customers start building alternatives. (TechCrunch)
The Onboard
This week's technique: custom slash commands. Codify a workflow you run every session into a single /name you never have to type out again.
Most operators know Claude Code has built-in slash commands. Fewer have built their own. A custom slash command is a markdown file in .claude/commands/ (project-scoped) or ~/.claude/commands/ (user-scoped). The filename becomes the command. The contents become the prompt Claude runs when you invoke it.
1. Create .claude/commands/review.md in your repo root. 2. Write your standard code-review prompt inside it, variables like $ARGUMENTS get replaced with whatever you type after the command. 3. In-session: type /review src/api/routes.py and Claude runs your full review prompt against that file.
You'll know it worked when you stop re-typing the same 200-word prompt every morning and just hit /review instead.
The Playbook
The move: pipe your CHANGELOG into Claude before every standup.
Most teams update their CHANGELOG manually, which means it's always stale. Here's the tighter version:
1. Run git log --oneline --since="24 hours ago" to pull the last day's commits. 2. Pipe the output to Claude with the prompt: "Summarize these commits as a plain-English CHANGELOG entry for a non-technical stakeholder. Group by feature area. Flag anything that looks like a breaking change." 3. Review the output in under a minute, paste the keepers into your CHANGELOG or Notion.
You'll know it worked when your 9am standup note writes itself and you stop explaining "what you shipped yesterday" from memory.
Works best when commit messages follow a consistent convention (feat/fix/chore prefixes). If yours don't, add a step: ask Claude to normalize the commit subjects first, then summarize.
Builder's Brief
We build The AIgent's engine in the open. An honest look at what we are making, what broke, and where it is headed.
We're heads-down in the workshop this week. The next dispatch from inside the build lands in the coming days, the honest version: what we shipped, what broke, and where it is headed. Thanks for following along.
Building something similar? Hit reply and tell us what you're working on. We read every one.
|
Recommended reading
If you like The AIgent, a small group of operator-tier publications worth your inbox: see the shortlist. |
Before You Go
Ten drops, an agent multiplexer with 13k stars, and Claude's now living in your pocket. The tools are moving faster than most operators can install them. The discipline is knowing which three to actually ship with.
See you Thursday.
