plockett.eth (🅰️,🌶)

💃I'm back to building developer tools! Please help me better understand your siiick dev setups with coding agents or your frictions/pains by completing this 3-5min survey: forms.gle/gdn6vNCRpWfwFf….

@Shpigford use browser harness and just login for it in local chrome.

what's the best SEO MCP/CLI tool? i don't want to dig through some complex UI. i want to hop in claude/hermes and just say "based on my product at replysocial.⁠co, what are 20 starting keywords i should focus on ranking for?"

@lateinteraction Look Omar. I just got local folks to understand the Mismanaged Geniuses Hypothesis, RLMs, and thin hanresses. I say all that to say... keep it coming !!

We're gearing up to release two research efforts I've been extremely excited about for quite some time. Y'all will really love these.

@VictorTaelin @honchodotdev Also add skills so it take a more constrained approach to work that has footguns like this.

@VictorTaelin The answer is @honchodotdev by the 2nd or 3rd time it will remind itself

seriously, working with AI is MISERABLE for one and only one reason: having to re-explain the same thing "oh yeah this new session obviously doesn't know what proper case trees are, so let me explain it for the 5000th time in my life" I'm tired AGENTS.md doesn't solve this because it is impossible to fit the entire domain knowledge without nuking the context - it would be 1m+ tokens worth RAGs don't solve this, the agent won't search unknown unknowns SKILLs don't solve this unless I keep like a collection of 1750 skills with specific cuts of domain knowledge for each possible subset of my domain that I might need in a given chat, but that's a lot of manual work recursive LLMs or whatever don't solve this for the same reason, you can't dump a domain book and expect the AGENT will magically guess that it is supposed to search for a specific bit knowledge. unknown unknowns fine tuning doesn't solve this (OSS models suck and OpenAI / Anthropic gave up on user fine tuning) I honestly think a good product around fine tuning on your domain would be a major hit and an underdog lab should take this opportunity

@therick @amasad @pirroh Have him join @UnlockAlabama !

@amasad @pirroh Awesome! 👏 My son is graduating from University of Alabama this week and this is exactly what he needs to supplement an Accounting/Finance degree.

Thrilled to support the education GOAT as he reinvents education (again).

Just completed the hard challenge that got me into CS with @browser_use browser harness + an additional harness I made that is similar to it. No other agent + browser was able to achieve what was needed reliably enough with enough ease till this one. Lots of obscure government data is about to get processed.

My Claude keeps saying "that's a day of work" when it's like done in 5 min 😂

@MiTiBennett I like that you and @lateinteraction make wild, groundbreaking research, but will come on x and make these fun, simple jokes.

this is awesome. i don't know how they got 6 children with one injection, but it's an amazing achievement

@ChrisLaubAI @madhavjha could you make a browser harness for this? the best one uses cdp

A Rust dev just killed Headless Chrome. It's called Obscura. The open-source headless browser purpose-built for AI agents and scrapers at scale. Chrome vs Obscura: - Memory: 200MB+ → 30MB - Binary: 300MB+ → 70MB - Page load: 500ms → 85ms - Startup: 2s → Instant - Anti-detect: None → Built-in Single binary. No Node, no Chrome, no dependencies. Stealth mode is brutal: → Per-session fingerprint randomization (GPU, canvas, audio, battery) → 3,520 tracker domains blocked by default → navigator.webdriver masked to match real Chrome → Native function masking so detectors can't sniff it out Drop-in replacement for Puppeteer and Playwright over CDP. Zero code changes. If you run agents or serious scraping at scale, this repo prints money. 100% Opensource.

Today, we're shining a spotlight on our intern Madison Zoebisch, a sophomore at Tuskegee University. Madison is building The Skegee Continuum, which is a centralized hub for student support. Students can ask the chatbot questions and receive quick, easy information on on-campus resources and on-campus information.

@gregpr07 @browser_use Btw @MiTiBennett has explained this plainly as Weak maxing. Been building harnesses with this principle too.

@gregpr07 @browser_use Hold my beer

Apply in our Bio!

@gregpr07 I will try it but since you asked. no browser agent I have found can do this task: download all files for a given year on this website -> app.adem.alabama.gov/eFile/

Introducing: Browser Harness. A self-healing harness that can complete virtually any browser task. ♞ We got tired of browser frameworks restricting the LLM. So we removed the framework. > Self-healing — edits helpers. py on the fly > Direct CDP — one websocket to Chrome > No framework, no rails, complete freedom > Drop-in for Claude Code and Codex I challenge anyone to find a task that DOESN'T work. I couldn't yet.🔥 100% open source ↓

@dosco @DSPyOSS gets even better if you add “hooks”, ie some validation that makes a bad “done()” look like a failed function call with a reason. so that it starts “debugging” to a correct its error.

very cool writeup on applying RLM and DSPy to multi-modal data. this bit really got me thinking...

To build the GitHub for robots, we must first build an all in one repository for your robotics project. We just did that. Try it now tnkr.ai

#Claude is not down anymore. Back to telling it what to do.

@dosco Activity using it to make an agent harness on a contract job now 🫡

would love more folks using and helping improve the DSPy based RLM+GEPA agent substrate 1. detailed design doc github.com/ax-llm/ax/blob… 2. claude skill to use it github.com/ax-llm/ax/blob…

agents out there are struggling with these naive and brittle implementations the paper barely touched the real key to unlock agents, RLM with GEPA optimization of the core agent + harness code with a solid eval harness.