genedai

vor 3 Monatendiscuss

That discomfort of letting an agent touch your personal inbox was actually the core insight for me. You lose visibility into what it's doing, and mixing agent actions with human mail is just a recipe for confusion. An agent should have its own identity, its own address, its own inbox. Would love to see where you're taking agentmail — are you focused more on the agent-as-sender side, or agent-as-receiver (verification codes, inbound parsing, etc.)?

genedai

vor 3 Monatendiscuss

I've been building AI agents for the past year, and I kept running into the same problem: my agent couldn't sign up for anything.

  It needed to create accounts, which means receiving verification
  codes. It needed to send reports. It needed to check inboxes. Every
  time, I'd hack something together. Resend for sending (but it can't
  receive). Gmail API with its OAuth setup that takes half an hour. Or
   just... doing it by hand.

  So I built mails-agent. The idea is simple. Give your agent a real
  email address, the same way you'd give it a tool or an API key.

    npm install -g mails-agent
    mails claim myagent
    mails code --to [email protected] --timeout 60

  That last command waits for a verification code to arrive and prints
   it. Your agent can now sign up for things on its own.

  I've spent years building products for people. This is the first
  time I'm building something for agents, and honestly it's the most
  fun I've had in a long time. It's a completely different mindset.
  You're not thinking about UI or attention spans. You're thinking
  about reliability, programmability, zero friction. Everything is an
  API call. It just needs to work.

  A couple of things I'm happy with:

  1. It runs entirely on Cloudflare's free tier (Workers + D1 + R2 +
  Email Routing). I wanted this to cost $0. Charging per mailbox felt
  wrong for something agents need as a basic capability.

  2. It works as an MCP server, so Claude Desktop and Cursor can use
  it directly: npx mails-agent-mcp

  MIT licensed. Self-hostable. Deploy to your own Cloudflare account
  with your own domain, or claim a free @mails0.com mailbox to try it.

  I'd love to hear what you'd use this for. I built it for the signup
  and verify use case, but I think there are agent email patterns I
  haven't thought of yet.

  Demo: https://mails0.com

genedai

vor 4 Monatendiscuss

Been using this for a few days, works really well. Way less token waste compared to the usual CDP approach.

genedai

vor 4 Monatendiscuss

Haha, “OpenJobs” here means open jobs / open opportunities, not that the core product is open source. Fair question though, since “open” is definitely overloaded in AI now.

genedai

vor 4 Monatendiscuss

[dead]

genedai

vor 4 Monatendiscuss

Look, I know everyone's freaking out about the MacBook Neo at $599. I get it. It's cheap, it's Apple, take my money. But honestly? If you spend like five minutes actually reading the specs, the MacBook Air is the obvious buy here. The Neo sounds sick until you look at what they actually shipped. It's running an A18 Pro. That's a phone chip. Not even an M-series. 8 gigs of RAM. One of the USB-C ports is USB 2.0, which, like... it's 2026, what are we doing. No MagSafe. No Thunderbolt, so forget plugging in a Studio Display. No backlit keyboard. No Force Touch. The base model doesn't even have Touch ID. Someone on 512pixels put together a list of everything they cut and it's honestly kind of depressing. Just cut after cut after cut. The Air at $1099 gets you the M5, 16 gigs of RAM, Thunderbolt 4, MagSafe, Wi-Fi 7, P3 display, backlit keyboard. The whole deal, no weird gotchas. Yeah it's 500 bucks more. But you're getting double the RAM, a way better chip, and you don't have to keep a mental list of stuff your laptop randomly can't do. The RAM thing is what really kills the Neo for me though. macOS eats like 5 gigs just sitting there doing nothing. Someone on HN said they're running VSCode, Xcode, Blender, and ChatGPT all at once on 8 gigs and sure, macOS is clever with swap and compression, but come on. You're redlining it from the jump with zero room to grow. And you can never upgrade it. 16 gigs on the Air means you're actually comfortable for years. This matters way more now because of the local AI stuff that's blowing up. OpenClaw just dropped, open-source AI agent that runs right on your Mac, handles your email, calendar, does actual useful stuff on its own. The LocalLLaMA crowd is going absolutely feral setting it up with Ollama, running local models for free. People are hitting 49 tokens per second, running 30B parameter models on their Macs. That kind of thing eats RAM for breakfast and really wants those M-series GPU cores and the Neural Engine. On 16 gigs with an M5? You're golden. On 8 gigs with a phone chip and a 5-core GPU? Yeah no. The move in 2026 isn't buying the cheapest Mac. It's buying the one that doesn't feel slow in two years. Grab a refurb M4 Air for like $900, or just pay full price for the M5. Either way you've got a machine that handles everything for the next five or six years easy. Browsing, dev work, local AI stuff, whatever. The Neo is gonna start choking the second you try anything beyond Safari and Netflix, and at that point just buy a Chromebook for 300 bucks. Apple built the Neo to hit a price tag. The Air is where the math actually works out. Spend the extra 500 now, forget about it for five years. Done.

genedai

vor 4 Monatendiscuss

I got tired of AI giving me "it depends" answers when I asked it product management questions. So I built a SKILL.md file -- pure Markdown, zero scripts, zero dependencies -- that turns Claude Code (or Cursor, Windsurf, Codex) into an opinionated PM agent with real domain knowledge.

The architecture is intentionally boring. A single SKILL.md file acts as a router: it maps 40+ user intents to 6 knowledge modules loaded on demand from a `knowledge/` directory. Each module contains frameworks, decision trees, quality gates, and anti-pattern detectors. There are 12 templates in a `templates/` directory. Total system: ~130KB across 20 Markdown files. No build step, no runtime, no API calls. You can read every line before installing.

What makes it different from just prompting "act like a PM": the knowledge modules encode actual domain logic. The finance module has 32 SaaS metrics with exact formulas and stage-specific benchmarks. Ask it about 8% monthly churn and it doesn't say "that's high" -- it computes `1 - (1 - 0.08)^12 = 63%` annual churn, flags it as a red severity indicator, and tells you the benchmark for your stage. The discovery module enforces Mom Test principles and catches "Solution Smuggling" (sneaking a solution into a problem statement). The artifacts module has 8 story splitting patterns and 9 epic breakdown patterns, not just a template to fill in.

The quality gate system is the part I'm most interested in feedback on. Every output passes through universal gates (assumptions must be labeled, outcomes must be measurable, no "Feature Factory" outputs) plus domain-specific gates loaded from the active knowledge module. The agent is designed to push back -- if you ask for a PRD without a clear problem statement, it will challenge you rather than generate filler.

Tradeoffs worth noting: it's CC BY-NC-SA 4.0, not MIT, since the knowledge modules represent significant curation work. It's also opinionated by design, which means it won't work well if you want a yes-machine. And it's optimized for Claude Code's SKILL.md convention -- it works in other editors that support Markdown skills but the routing system is tuned for Claude's context window management.

GitHub: https://github.com/Digidai/product-manager-skills

Install: `clawhub install product-manager-skills`

genedai

vor 6 Monatendiscuss

[dead]

Submissions

[untitled]

[untitled]

Give your AI agent a real email address (open source, Cloudflare)

Show HN: OpenJobs AI – An AI agent for outbound recruiting and sourcing

Show HN: JD Roast – Paste a job description, get it brutally roasted

A 130KB Markdown file that turns Claude Code into an opinionated senior PM

comments