In partnership with
The job search repo on today's trending page is the one worth sitting with. Not because of the stars, because of the shape of it: a geophysicist got cut, built a tool to run his own search, and it got him hired. That's the whole pitch for every repo below. Nobody is waiting for permission to ship something useful anymore, they're just doing it and open-sourcing the receipts.
|
Sponsored
|
Why Most AI Tools Fail at B2B Customer Service
B2B customer issues rarely resolve in a single interaction. They move across teams, systems, and account history that most AI tools were never designed to track.
A new briefing paper from Harvard Business Review Analytic Services, sponsored by Front, looks at why AI built for high-volume consumer support creates new coordination gaps when it meets the multi-team reality of B2B service, and what leaders should ask before investing in their next AI tool.
Get the free briefing paper to see the three questions that separate tools built for your environment from tools that will create more work.
The Drops
Repoopencode
216,404 stars · anomalyco/opencode
The open source AI coding agent. It works with 75+ model providers, local models included, so no single vendor owns your workflow.
Here is the real problem it solves. Every hosted coding agent ties you to its pricing, its uptime, and its model roadmap. This one doesn't.
Quick start: curl -fsSL https://opencode.ai/v2/install | bash
Catch: the desktop app is still in beta, so the CLI is the stable path for now.
AffiliateHubSpot
Picking your own coding agent is one way to stop renting your workflow. If you also run customer work, HubSpot is a CRM with prebuilt AI agents that run on the customer record. A Customer Agent answers support, a Prospecting Agent researches and drafts outreach, a Data Agent cleans the records. The CRM underneath has a free tier; the agents are add-ons.
Use it for: putting support, prospecting and data-cleanup agents on the record your team already keeps.
We may earn a commission.
Repoai-job-search
52,598 stars · MadsLorentzen/ai-job-search
An AI job application framework built on Claude Code that evaluates postings, tailors your CV, writes cover letters, and preps interviews.
Use it for: running your own job search the way the author did, who turned 69 tailored applications into a signed contract.
Catch: the job board search skills ship for Danish portals; anywhere else you add your own board with /add-portal.
Repofreehire
1,227 stars · strelov1/freehire
An open-source job search engine pulling postings straight from company ATS platforms, Workday, Greenhouse, Lever, and more.
Use it for: a job feed with no recruiter reposts and no dead links, self-hostable if you want your own copy.
Catch: none stated in the README.
MCPSolidCraft
5,388 stars · tangency-labs/solidcraft
An open-source parametric 3D CAD app built in pure Rust, with a built-in MCP server so agents can model through it directly.
Use it for: letting an agent drive actual CAD work, changing a dimension and watching the model rebuild.
Catch: the README calls it early but usable for simple parts.
|
From Our Partners
|
Leave Granola and get up to 12 months free of Wispr Flow Notetaker + Dictation
If you have paid time left on an individual Granola plan, we'll match it with a Wispr Flow subscription that includes Notetaker and dictation, and add bonus time, up to 12 months total. Sign in or create a Wispr account and submit proof of your plan to check eligibility.
Start Here
Turn your one-sentence plan into a two-week schedule a team can actually follow.
One prompt today instead of a how-to. It is one of ours, tested on ChatGPT and Claude. Copy it, paste it into the free app you already use, and fill in your list of tasks, who is doing the work, and any real constraints you're working around.
Tasks:
[TASK LIST]
Who is doing the work: [TEAM SIZE AND COMPOSITION]
Constraints: [CONSTRAINTS]
Give me:
1. The goal for the two weeks in one sentence: what is true at the end that is not true now.
2. The tasks in the order they should be done, split into week one and week two. For each one, use supplied effort estimates or label your own as assumptions: under a day, a few days, or most of a week.
3. Anything that has to wait on something else finishing first, and what it is waiting on.
4. What gets cut if this is more work than two weeks holds. Do not hand me options to choose between. Name any tasks you recommend cutting and why. If everything reasonably fits the stated capacity, say so; do not invent cuts.
If I left who is doing the work or my constraints blank, ask me for them before you build the plan. If you build it anyway, do not invent a team, a deadline or a cut, and label every assumption you made as your own.
The line that makes it work: "What gets cut if this is more work than two weeks holds... Do not hand me options to choose between. Name any tasks you recommend cutting and why." Most planning prompts stop at listing tasks. This one forces the model to make the hard call about what doesn't fit, instead of handing you a wishlist disguised as a plan.
TRY THIS: open the notes app where your task list already lives, copy it as is, and paste it straight into this prompt. Don't clean it up first, let the model sort the mess.
This is one of 100 in the AIgent prompt pack. The other 99, plus guided builds with Mike on video, are inside the AIgent membership: Join the AIgent membership. Seven days free, then $29 a month. One click to stop.
|
Recommended
Once you've got a plan a team can follow, the next bottleneck is usually the newsletter or update that reports on it, which is exactly the kind of writing you can hand off entirely. HeyNews, the AI newsletter writer that sounds like you learns your voice from past issues, pulls stories from your sources, and hands you a publish-ready draft in minutes; 14-day free trial, then 50% off for 12 months with code WELCOME50. We may earn a commission. |
Frontier Signals
OpenAI's own models hacked a crypto wallet and breached websites during testing. Four new incidents from June, disclosed now: agents that hit access denials broke past a site's authorization checks, and one used funds from a locked wallet on a test network. If you're giving an agent real credentials, assume it will use them the way a very literal intern would: exactly as far as the guardrail lets it. (Ars Technica)
Vercel's AI Gateway now logs session affinity, showing which app ran which session and what it cost. Claude Code, Codex, and Cursor sessions are now individually traceable with cost attached. If you're running multiple coding agents against one gateway, this is the first real way to see which one is burning your budget. (Vercel)
Meta's AI agent Muse is allowed to sext with adult users, while adult creators say their own accounts get restricted. Meta says a private chat with an AI is different from promoting content on a social platform; creators see a double standard. Worth watching if you build anything that touches platform content policy: the rules can apply differently to the house's own bot. (Wired)
Disney is set to hire most of Character.AI's workforce in what's structured as a stalking horse bid. Another company could still outbid Disney for the deal. Worth noting less for the acquisition itself than for what it says about where character-agent teams end up when the standalone business doesn't work. (TechCrunch)
|
Recommended reading
If you like The AIgent, a small group of operator-tier publications worth your inbox: see the shortlist. |
What do you spend on AI tools in a typical month?
Before You Go
Reply and tell me which of today's repos you're actually going to clone this week. I read every one.
See you Tuesday.
Before you go: we started a room for people actually building with agents. Stuck on something? Post it in General in the community and we'll help you get unstuck. Join the community →
|
Want to reach builders shipping with AI every weekday? Advertise with The AIgent. The AIgent stays free for everyone, and reader support keeps it that way: Support The AIgent. |
