PH热榜 | 2026-07-19

finally, an open-source alternative to @Ahrefs and @SEMrush. it's about time.

S/O to @bensenescu, keep up the great work. oss ftw!

@bensenescu
Glad someone finally built this. A lot of small teams give up on SEO before they start, because the tools cost $100/mo before you've written a single post.

So the pricing is what caught my eye. Backlink and keyword data is usually the expensive part — how are you sourcing it? Own crawler, or licensing an index?

The MCP angle is interesting too. We work on the content planning side, and what we keep seeing is that agents will build a whole strategy on bad keyword data and sound completely confident about it. Do people trust the output more when the numbers are real, or does it just move the guessing one step later?

Starred the repo. Hope this one does well.

Hello 👋

This February, I was building other open source projects and wanted to do SEO.

I was floored by how expensive and bloated the tools were. There weren't any modern open source solutions, so I figured I should just build that.

Since then, it's been really cool seeing founders and agencies enjoy the product and even building their own custom features.

Hope you give it a try: openseo.so. There are some trial credits to test it out, then it starts at $10/month.

You can also self host on Cloudflare or with Docker.

Give us a star on github or you can join the discord to get SEO advice, contribute, or just hang out.

One thing that would make this way more useful for me is letting the MCP agent pull historical ranking data so I can see if keyword positions are actually moving over time, not just a snapshot of what the competitor looks like today.

This is gunna be huge. Well done

On the content-brief idea you're chewing on with @resulolys: I'd resist baking a fixed brief format into the tool at all. When we wired structured data into our agents, the fixed-template version kept fighting us the second someone wanted a different outline, and blogs vs landing pages vs comparison pages each want a different scaffold. Exposing keyword, competitor, and backlink as separate MCP tools and letting the agent compose the brief (your Hermes/self-updating-skills instinct) ages better than one generator trying to cover every type. One watch-out with DataForSEO behind MCP: the agent will treat those volumes as ground truth, so stamping an as-of date on each metric keeps it from acting on a stale number.

open source is the right call here for something this data-heavy - keeps you from getting locked into one vendor's index. curious how you're sourcing backlink data at that price point though, is it your own crawler or are you licensing an existing index and just eating the margin?

Can it be the better tool than ahrefs

The $10 vs $100+ framing is what got me. I've watched so many small teams skip SEO tools entirely because the pricing assumes you're an agency, so an open source option at a tenth of the cost fills a real gap.

This is a great price for independent developers. Adopting the MCP method to integrate into my Agent is much smarter than adding another platform.

honestly the price point is what caught my eye first, like $10 is kind of a no-brainer compared to what i was paying before. the MCP connection actually works smoothly with my setup which surprised me.

would be awesome if you could add a content brief generator that pulls keyword and competitor data straight from the dashboard into a doc template, basically saving the back and forth between the tool and writing apps

honestly the MCP integration is a really smart move, makes it feel like the tool actually fits into how people already work with their agents instead of being yet another standalone dashboard

Love the pricing and the MCP angle is really smart. One thing that would help me as a user is a built-in content brief generator that pulls the keyword, competitor, and backlink data together into a single outline my agent can act on. Right now I still end up stitching the pieces together myself in a doc.

Love how clean the MCP integration setup is. Most SEO tools bury that kind of connection behind ten menus, but here it feels like a one-click setup that actually respects how someone using an AI agent wants to work.

Congrats on the launch. The MCP plus real SEO data angle is strong. The part I would want as a user is traceability: when an agent suggests a keyword or page update, can I click through to the actual SERP, keyword, or backlink evidence it used? That would make agent SEO recommendations much easier to trust.

@bensenescu The GSC loop is the part I'd stress-test. When we wired an agent to the Search Console API, the last 2-3 days of rows were still settling, so anything published Friday read as flat until Monday and the agent kept trying to "fix" work that was actually fine. We ended up throwing away the trailing 72 hours before letting it draw any conclusion. Does yours hold that window back, or does it read the freshest rows it can get?

@Ben Senescu makes sense, DataForSEO is a solid choice, that keeps you from having to run your own crawler at that scale. good answer, thanks.

I've been on the content side for a while — writing-focused, not SEO-first — and the pricing of every serious SEO tool has always assumed you're an agency running five client accounts. You just need to know which topic gaps actually exist before you spend three weeks writing something nobody will find.

The open-source angle here matters more than the price tag, though. Most content people I know aren't going to self-host anything, but knowing the data layer is DataForSEO and auditable changes how much you trust the keyword numbers. Black-box confidence scores are the part that always made me discount the recommendations anyway.

Congrats on shipping this — curious how the content-brief workflow compares to using the backlink data directly for topic research rather than competitive analysis.

This fills an immediate need and fits right into my agent workflow. Trying it out right now.

The $10/mo + MCP + GSC loop is a genuinely appealing combo for solo founders who can't justify a $100 Ahrefs seat. Curious about the cold-start case though: for a brand-new domain with basically no rankings or backlinks yet, does the agent still give useful direction, or does it need some baseline history to work from first?

The data grounding solves half the problem. The other half is what the agent does with it: give an agent keyword targets and it will happily over optimize, stuffing the exact phrases until the content reads generic and obviously machine written. Real data plus a light touch on the writing is where this gets genuinely useful. How opinionated is it about the output, or does it leave the writing style to you?

Thanks for the straight answers on DataForSEO all through this thread. One follow-up now that GSC is in the loop: does the agent get ranking history as its own MCP tool, or does it come back bundled with the keyword call? We found agents behave very differently when the "what happened after we shipped" data is a separate tool they have to decide to call, instead of something that arrives for free with the research.

Congrats — #1 on a Sunday is no accident. The self-host option is the part I find most interesting: if I run it on my own Cloudflare/Docker, what happens to the data layer? Do I bring my own DataForSEO key and pay them directly, or does usage still route through your account? Asking because that's usually where 'open source' SEO tools quietly stop being self-sufficient — the index is the moat, not the code.

This looks and work really well, just 10 minutes in using it and I can see the ease-of-use and clarity! Adding my 2nd project and will add more. Tip: when adding 1st project it didn't pick the name of the "google search console", just default, so it was a bit tricky to figure out that "a project" was the way to add another. Great work otherwise!

Open-source SEO is overdue, and the MCP angle is smart — letting an agent actually pull the data instead of guessing is the whole point. The hard part of any Ahrefs alt has always been the backlink index; that's the real moat, not the UI. Where does your link/keyword data come from, and how fresh is it? Following.

what's the data source behind it ?

Hi I wish there's more guidance on how to use it properly. Where can I find a tutorial on best practices to use this?

the price point alone made me try it, and the MCP integration actually worked smoothly with my agent. keyword data feels solid so far, will keep poking at it.

real data behind the mcp is the smart part, most seo agents just vibe keywords. does the agent loop back to see which ones moved rankings after publish?

the duck-as-a-spy branding is doing a lot of work here, it's memorable. question on the price history itself - when I add a product I just found today, do you backfill history from somewhere, or does the chart only start from the moment I add it, so the "is this actually a good deal" signal takes a while to become useful?

This is a very relatable problem. The product seems strongest if it turns “I bought too early” into a simple post-purchase routine rather than another tracker to manage. Do you monitor price drops automatically from receipts or orders, or does the user manually add each purchase?

I don't even trust discounts these days for this exact reason. Will try this out!

Context specific is the right call. What saved us when we stored scraped prices was writing the request context into the same row as the number, so region, currency and whether we were logged in all travelled with it. Before that we lost an afternoon chasing a price drop that turned out to be a different geo answering the same URL.

Love the idea, I've been burned by fake discounts before, but I also wonder how often the best price is worth waiting weeks or months for.

Congrats on the launch, Hlib.

Ran a quick manual QA pass over the site out of habit. One thing worth

checking right now: your Pricing section stays on "Loading available

plans..." and the plan data isn't in the served HTML, so the Pricing

link in your header currently leads to a section with no prices.

Found a few other bits too. Happy to send the write-up over if useful -

free, no strings.

The duck-spy branding got me before I even finished the tagline, nice touch. The thing I kept hitting when I built price scrapers is that a lot of retailers personalize or geo-vary the price now, and some quietly A/B test it mid-session, so 'the price yesterday' isn't one number, it's a spread. Do you snapshot from a fixed region and session so the history stays apples-to-apples, or could two people see different Spycost baselines for the same product? That's the part that decides whether the 'is this a real deal' call actually holds up over time.

Finally something that actually calls out fake discounts! So tired of "limited

time" sales that are always running. This is the kind of tool that makes

shopping way less stressful.

Strong pain point, beautiful solution, and a great launch :)
Good job, amigo.

Next Feature Idea: call bullshit on "only X seats left, etc" :P

Price history tracking should be built into every e-commerce site but it never will be because it's bad for business.

this is a real problem, fake discounts are everywhere. genuinely curious about the mechanics though - to know a price history you'd need to have been tracking a product before the "discount" showed up. does it only work for stores/products already in your database, or is there a way to check a price you're seeing for the first time?

Congrats on the launch, Hlib — the spy-duck is perfect for this. One thing I’m genuinely curious about: when I add a product, do I see its price history from before I started tracking (so I can judge today’s “deal” right away), or does the history only build up from the moment I add it?

Hey @Spycost Team, congrats on launching!

This resonates with me. Last November, I waited weeks for a deal on a coffee machine. When the discount finally appeared, something seemed off. I checked a screenshot I had taken earlier, and the "sale price" was the same as it had been a month earlier. They had increased the price just to reduce it again. I really felt manipulated. I wish I had Spycost then. I'm definitely going to try your tool.

Congrats on the launch! Is it capable of tracking a product CLASS, not just exact single product? A good example would be buying a coffee machine. You know a set of features/parameters you need, but you are open to look at multiple brands.

Would love a browser extension version so it can auto-track prices while I shop instead of me having to look things up manually every time, would make it way easier to catch those sneaky price jumps in the moment.

One thing that would make Spycost really useful for me is a simple browser extension that automatically flags price drops and fake discounts while I shop, instead of me having to look things up manually each time.

This is painfully relatable. a discount can make something feel urgent even when it was never a priority in the first place, and finding out later that the "sale" price was basically normal makes it even worse :)

I like that Spycost does not try to decide for the user, it just shows the real price history so the choice is based on context instead of pressure. Curious how it handles stores that constantly change product URLs, variants, or create fake reference prices to make discounts look bigger.

I had to add this product into my "Interesting" collection because I love discounts :D

Is there a browser extension that allow me to see price history for specific product directly on PDP?

Project look interesting, good luck 🍀 with launch

Hey Product Hunt! 👋

I'm Michael, the maker of Kobbe.

I built Kobbe because I wanted a web analytics tool that was simple, fast, privacy-friendly, and didn't overwhelm me with dashboards full of metrics I'd never use.

Launch offer: Start with a 15-day free trial, no credit card. When you upgrade to a yearly plan, use code hellokobbe for 20% off.

→ https://kobbe.io

With Kobbe you can:

• 📊 Track visitors, pages, referrers, devices, countries, and traffic channels

• 💰 Measure revenue attribution and conversions (Stripe, Polar, Paddle, Creem)

• 🔍 Connect Google Search Console for search terms and landing pages

• 🤖 Break out AI referrals (ChatGPT, Perplexity, Claude, and more)

• 🎯 Set up custom events, conversions, and funnels

• 🧩 Embed analytics widgets directly into your website

• 🚫 Privacy-friendly analytics without cookies

• ⚡ Install in minutes with a lightweight script

Whether you're building a SaaS, indie project, blog, or online business, Kobbe helps you understand what's working without the complexity of traditional analytics platforms.

Thanks for checking out Kobbe! 🚀

The revenue attribution is what stands out to me here. Most privacy-friendly analytics stop at pageviews, and then you can never show which content actually made money.

Two questions from the content side. How do you handle organic search as a source? Without cookies you usually just get "google.com" and no idea which page or query brought them. And do you separate AI referrals — ChatGPT, Perplexity, and so on? A lot of that traffic arrives with no referrer, and right now it's the thing people most want to see and mostly can't.

Also the dashboards look clean, which is rare in this category. Nice work.

Hey Michael! Love to see you behind this amazing product. Privacy is becoming more and more important than ever and Kobbe is facing it like nothing before. Wish you all the best here

Love how clean and privacy-first this looks, exactly what smaller sites need without all the consent banner hassle. One thing that would make it even better: a simple way to compare traffic and revenue side by side across date ranges, like a lightweight view that shows whether a traffic spike actually translated into sales. Would really help with quick decision making.

Finally tried Kobbe this week and was surprised how fast the setup was, like literally a copy-paste snippet and I had real funnels showing up. Love that I can finally drop GA without feeling like I’m going in blind on my own traffic.

I've tried a lot of analytics platforms but most of them were bloated and boring.

Kobbe is the first platform which is built on simplicity and design aesthetics in my experience and of course privacy.

Congratulations on the launch @michael_andreuzza, wishing you loads of success with Kobbe!!

Finally an analytics tool that doesn't need a cookie banner. Set it up in like two minutes and the dashboard is honestly cleaner than what I'm using now.

Love how clean the dashboard feels without sacrificing depth, especially seeing revenue and funnels tied together right next to traffic sources. The cookie-less setup is such a relief compared to wrestling with consent banners.

A heatmap view of where users actually click and scroll on each page would round this out nicely, especially if it stays on-device like the rest of your approach. That would make it a stronger competitor to tools people currently use alongside their analytics.

the company-identification feature is the interesting tension here - most "privacy-friendly" analytics tools stop at not using cookies, but reverse-IP company lookup is its own kind of tracking even without a cookie involved. is that opt-in per site, and do you disclose it to end visitors anywhere?

Privacy-friendly analytics with revenue attached is a useful angle. A lot of lightweight analytics tools stop at page views, while revenue attribution is where founders actually make decisions. How are you handling attribution when payment events happen in Stripe or another backend instead of on the marketing site?

cookie-less but still tracking revenue and funnels is the interesting combo here - how do you attribute a purchase back to a specific source/session without a persistent identifier? some kind of short-lived fingerprint hash that resets often enough to stay privacy friendly?

The revenue attribution piece is what I've been looking for in a privacy-friendly tool for a while. Most GA alternatives stop at pageviews and maybe events — you can see that a post got 2,000 readers, but there's no clean path to "and three of them became paying customers starting from this paragraph."

The Stripe/Polar/Paddle integrations make that loop actually closable for indie projects. Plausible gets close but it requires more custom event setup to do the same thing.

One thing I'd love to see: whether the funnel view can start from a specific page rather than only from the first touchpoint. For content-heavy sites the "which piece of writing converted someone" question is different from the "what was their first session" question.

Cookie-less plus revenue tracking is the interesting combo — attribution is exactly where cookie-less tools usually give up. How do funnels work across sessions without an identifier? Is it same-session only, or do you do some kind of privacy-preserving linking on the server side? If you've solved multi-session funnels without cookies, that's the headline IMO.

I think the problem in this space is convincing people to switch once they already have analytics installed. What has been the biggest reason users migrate to Kobbe?

"Attribute revenue without turning Kobbe into a CRM" is the line that got me — most privacy-first tools nail the traffic but lose the money trail, which is the part that actually matters to a small business. How do you tie revenue to a source without the cookies/cross-site tracking that make attribution "easy"? Looks clean.

Would love to see a heatmap integration that still respects the privacy stance, something click or scroll depth without recording session replays, since that would round out the funnel data nicely.

curious how much of this gain is specific to M5's tensor cores vs the general Metal 4 tensor API. I'm running an M4 Pro, plenty of people doing local inference haven't upgraded yet - does BaseRT still beat llama.cpp/MLX meaningfully on M4, just with a smaller multiplier, or is most of the win locked to the new hardware?

Installed it on my M2 and pulled a 7B model in under a minute, felt snappy even on battery. Nice to skip the per-token bill without cooking my laptop.

curious how it handles model swapping on the fly, like dropping into a bigger LLM only when a prompt clearly needs it. A simple toggle or auto-routing between something like Llama 3 8B and 70B based on the task would be a really nice touch for battery life on MacBooks.

One thing I'd love to see is a simple built-in model downloader so I can grab and switch between popular open weights from the CLI without hunting for GGUF files manually. That would make testing new models way smoother.

Love how simple the install is and the speed on my M2 is genuinely impressive. One thing that would make this even better is a built-in model benchmark tool so you can see tokens/sec for different models on your specific hardware right from the CLI.

honestly the install was just one command and my m2 was running llama models in like a minute, kind of wild how fast it feels compared to ollama

The one-command install is such a thoughtful touch, especially for folks who just want to experiment without dealing with setup headaches. Nice work making local LLMs feel that accessible on Apple Silicon.

Running Llama on my M2 was genuinely instant, no setup headaches. Loving that it just works without Docker or weird config files.

The speed numbers are impressive, especially on Apple hardware. For teams deciding between llama.cpp, MLX, and BaseRT, what is the tradeoff in model coverage or quantization support? I would love to know where you see BaseRT fitting in a production local-inference stack.

the Metal 4 tensor core optimization is the interesting part here. most local inference tools treat Apple Silicon as a nice-to-have but you're actually building for it as the primary target. curious how the performance scales with context length — prefill speed is great but does it hold up at 32k+ tokens?

The discussion's focused heavily on prefill vs. decode, but there's a third axis worth thinking about for Apple Silicon specifically: unified memory's effect on how many models you can keep resident simultaneously. llama.cpp and MLX both require fairly explicit load/unload, and on a machine where you might want a small coding model, a larger reasoning model, and an embedding model all warm at once, the round-trip cost of loading from disk breaks the local-first experience more than raw token throughput does. Is BaseRT doing anything with model-weight sharing or partial preloading, or is that a separate problem from what you've optimized?

On-device inference is exactly the direction I keep hoping more indie apps can go — no per-token bill and privacy by default is a big deal. Is BaseRT meant to be embedded inside a shipping Mac/iOS app, or is it more of a dev/CLI runtime for now? That'd decide whether I could actually bundle it into my own app.

Is it also on huggingface? I don't want to install it via bash command.

6.4x over llama.cpp is a big claim for a field this optimized — where does the speedup actually come from? Custom Metal kernels, better KV-cache layout, speculative decoding, or quantization formats? And does the gap hold at batch-1 decode on long contexts, or is it mostly a prefill win? Genuinely curious — we run local models for dev work and llama.cpp has been the default for so long that a 6x claim deserves a look.

6.4x faster than llama.cpp is a big claim — what's the tradeoff? Curious whether it's fast enough to run a real-time voice pipeline on-device (where every 100ms of latency is felt), or if the win is mainly on batch/throughput. Running local with no token cost + full privacy is a huge deal for small-business use cases. Following.

Very interesting work. It feels like we're finally seeing software catch up to what Apple Silicon has been capable of for a while.

One thing I'm curious about: if an application needs to switch between different models depending on latency or quality requirements, does BaseRT help manage that orchestration, or is that intentionally left to the application layer?

the no-per-token part matters as much as the speed imo, it's what makes always-on background agents affordable to run. is that the direction you're headed?

the prefill numbers are impressive, but prefill isn't usually where I feel the wait on a laptop - it's token-by-token decode speed during a long generation. is the decode-side speedup in the same range as prefill, or is that gap smaller since decode is more memory-bandwidth bound than compute bound?

@lukas_base Benchmarks you posted seems to be prefill numbers. do you have decode or token generation throughput comparisons against llama.cpp and MLX too, or is most of the gain concentrated in prompt processing rather than generation speed? Thanks!

Tuning specifically for the M5's new tensor cores through the Metal 4 API, instead of just optimizing existing kernels, is a genuinely sharp bet given how far the hardware has pulled ahead of the software stack. Your published numbers highlight prefill speed specifically, so how much does decode/token-generation throughput improve on M5 versus llama.cpp and MLX, since that's usually the bottleneck for longer, more conversational or agentic workloads?

@lukas_base Just a quick question how does it compare to LiteRT? It is no secret that MLX and llama.cpp are falling behind but we already have a great architecture from Google.

Took a few minutes to set up on my M2 and the speed genuinely surprised me compared to other runtimes I've tried. Love that everything stays on device, no tokens flying off to some server.

Keeping everything on-device and never saving screenshots is the smart trust move here, a memory tool only works if you feel safe letting it watch.

Screenshots-never-saved plus everything staying local answers the privacy objection most people would raise first. Curious about the SQLite file itself though: is FileVault the only thing standing between it and a plaintext read, or is there another layer on top?

Really exciting to see Rewisp — an ambient memory for your Mac live today. Big congratulations to everyone behind it! 🚀🚀

Would love a way to mark something as "do not track" when I'm working on sensitive stuff like medical forms or legal docs. A quick keyboard shortcut to pause Rewisp for 15 or 30 minutes would feel really natural and keep the privacy promise even stronger.

The semantic search feels surprisingly accurate for just text matching, and catching commitments like "I'll send it Friday" actually held up across a few days. The nightly summary at 9 PM is a nice touch without being noisy.

Love that it stays on-device and never saves screenshots. One thing I'd really appreciate is a way to exclude specific apps or windows from being read, like when I'm in a video call or working on something sensitive but unrelated to what I want tracked. That kind of granular control would make this feel truly mine.

The on-device text parsing with no screenshot storage is such a thoughtful privacy choice. Love that they figured out how to make it genuinely useful without crossing that line.

The promise-catching feature is probably the most useful detail here. small commitments like "I'll send it Friday" are exactly the kind of thing that disappears between chats, tabs, and notes, even though they often matter more than the tasks we formally write down.

I also like the privacy model: extracting text, immediately discarding screenshots, and keeping everything in one local SQLite file feels much easier to trust. Curious how Rewisp distinguishes a real commitment from casual language or quoted text without creating too many false reminders :)

Caught me the moment I told someone I'd send a draft Friday. Searched by meaning and actually found that Slack thread from last week. Quiet little tool, feels right at home on the menu bar.

the promise-catching and "what changed since Tuesday" bits are the sell for me over a plain screenshot recorder. one solo-dev question: months of full screen text in one SQLite file, does that stay fast to search, or do you end up needing to prune/summarize older entries once it's been running a while?

The ambient memory idea feels useful for people who context-switch all day. The tricky part for me would be separating “I saw this once” from “this is worth remembering.” Are you doing any local filtering or recency scoring so old or accidental screen content does not pollute answers?

on-device is the right call for something reading everything on your screen all day, but on-device isn't automatically safe - if my Mac gets stolen or someone gets local access, is that stored text file encrypted at rest, or just sitting in plaintext waiting to be read?

Congrats on the launch and #5th rank on Sunday.
Love the promise-catching idea. that's the stuff that always slips through the cracks. Does it handle commitments made in voice memos or dictated text, or only typed content?

As a fellow solo dev, the privacy-by-construction bit (text-only, screenshots discarded on the spot) is what makes this trustworthy vs a plain screen recorder. One thing I always hit shipping Mac apps though — the screen-recording permission prompt on first launch scares a chunk of users off. How are you handling that onboarding moment so people don't bounce before they get to the value?

What does the always-on screen reading do to battery and CPU on a laptop? I like the idea of asking "what changed since Tuesday" but I'd want to know the cost of having it watch everything all day.

The "what changed since Tuesday" search actually pulled up stuff I'd completely forgotten about scrolling through, which was a nice surprise. The on-device-only approach makes me feel way better about letting it read everything on my screen.

love that this exists, the filtering sounds solid. one thing i'd actually want though is a "near me" mode that uses your location to surface places within a driving radius, kind of like setting a max distance so you can find hidden spots for a weekend trip without manually scanning the map

So cool! Awesome work collecting this data. I’m from Brazil, but I lived in the Czech Republic for a while and had the chance to visit many of the “hidden gems” that most people wouldn’t find on Google Maps, simply because I made local friends who took me to those places.

I looked into it, and Detourmap is really solid when it comes to finding places that locals don’t usually tell everyone about (at least in Europe).

To be honest, I noticed a gap in my area of Brazil. I live in a small state called Espírito Santo, where we have countless waterfalls and hidden spots to explore, but the current data didn’t find many of them. I can definitely see how challenging it is to collect such a specific dataset. Have you considered encouraging community contributions so people can submit new places?

70k places is honestly a lot more than I expected, and the filter setup makes it really easy to just zoom in on something specific like castles or waterfalls without scrolling forever.

Love how dense the map feels without being overwhelming, and the filter for stuff like sacred sites and ghost towns is a really nice touch for planning weirder trips.

The fame filter running from world-famous down to deep-cut hidden gems is the part I'd actually use — most travel maps just drown you in the obvious stuff. When I flip on a few layers and set the fame filter for, say, caves and ruins in Portugal, can I share that exact filtered view as a link so a travel buddy opens the same map state? And is there a way to save pins into a personal list I can pull back up on the trip, or export them to Google Maps for navigation once I'm actually on the road?

70k pins across 350 countries is a great scope for this. one thing I'd wonder about with a Wikidata-sourced map - coverage tends to skew toward regions with a lot of editors, so western europe ends up way denser than say central asia or west africa even if those places have just as much worth seeing. does the density on the map reflect that editor bias, or have you done anything to correct for it?

This looks superb! What is the best way to suggest hidden gems? Could this integrate with All Trails?

Just spent a bit flipping the layers on the globe — the taxonomy is what got me (~14k sacred sites vs. under 600 ghost towns is a great contrast). One question: how’s “fame” scored for the famous → hidden-gem filter — Wikipedia sitelinks, pageviews, or something of your own? Congrats on the launch!

Finally a travel map that doesn't just slap pins on cities. Filtering by waterfalls and ruins is super satisfying.

Really cool concept, the filters are a nice touch for narrowing things down. One thing that would help me actually use it on the road is offline access or downloadable region packs, since signal can get spotty when you're already out of your way at a remote ruin or waterfall.

Finally a map tool that feels built for actual daydreaming. I flipped on castles and got completely lost for twenty minutes wandering through Romania and Scotland.

70k places is wild, and the country filter actually makes it usable instead of overwhelming. Spent a few minutes hunting for waterfalls in iceland and found three i'd never heard of.

Finally a map that actually makes me want to get off the beaten path, and the filter system makes it so easy to hunt down hidden waterfalls and ghost towns near wherever I am traveling.

Would love to see a road trip mode that lets you draw a route between two points and shows every detour-worthy spot along the way, sorted by how little extra driving each one adds.

love the "surprise me" bias toward obscure over famous, that's the right default for this kind of tool. since it's built off Wikidata, how do you handle places that get edited or delisted after you pull the data - is there a refresh cycle, or does it slowly drift out of date until you re-sync?

I'm Arshita, founder of CitedSpy.

A few months ago, I asked ChatGPT for the best tools in a category we were serving. Our company wasn't mentioned. That sent me down a rabbit hole.

I realized buyers are no longer discovering products only through Google. They're asking ChatGPT, Gemini, Claude, and Perplexity. Yet most brands have no idea what AI is saying about them, who gets recommended, or which sources influence those recommendations.

So we built CitedSpy.

CitedSpy helps brands track their visibility across AI search engines, monitor citations and sentiment, benchmark competitors, and understand how they appear in AI-generated answers.

We're excited to launch today and would love your feedback.

What's the most surprising thing you've seen an AI assistant say about your brand or industry? 👇

— Arshita Sharma, Founder @ CitedSpy 🚀

Congrats on the launch! Does it actually track based on different locations? I know a few brands and a few prompts that respond differently in each location, like the same prompt responds differently in the USA and India. Does your product support this?

honestly the cross-platform tracking across all the major AI assistants in one dashboard is super clean, you know most tools only cover one or two and make you piece the rest together yourself

Finally, we're live on Product Hunt! 🚀

After months of building, countless conversations, and lots of iterations, CitedSpy is officially out in the world. We'd genuinely appreciate your support an upvote, comment, or share goes a long way in helping us reach more founders and marketers.

Thank you for being part of our journey! ❤️

Needed product nowadays, congrats on the launch!

The note about turning down paid upvotes is honestly one of the strongest parts of this launch. building a product around authentic visibility, then refusing to manufacture that visibility on PH, makes the positioning much easier to trust :)

I've been thinking more about AI discoverability while building a product and its documentation, and the hardest part seems to be separating real visibility trends from normal answer variation. Curious how CitedSpy handles that across different models, prompts, locations, and repeated runs, and how it decides when a citation gap is actually meaningful enough to act on.

Really exciting to see CitedSpy live today. Big congratulations to everyone behind it! 🚀🚀

finally a tool that shows how often my brand shows up in ChatGPT and Perplexity responses, and the competitor comparison actually made me rethink our content priorities

How stable are these rankings ? If I found out my competitor is recommended more often, what can I do to improve that ?

A lot of people asked us:

"How does CitedSpy compare with other GEO tools?"

So instead of answering one by one, we built a dedicated Alternatives hub.

Compare CitedSpy with the leading AI visibility platforms, understand the differences, and choose what fits your needs whether it's us or someone else.

Explore here: https://www.citedspy.com/alternatives

the measurement part makes sense, but I'm stuck on the action part - once you know your citation rate is low, what's the actual lever to pull? unlike classic SEO there's no sitemap to submit to an LLM, so is the fix really just "publish more content and hope it gets into the training/retrieval data", or is there something more direct you can do?

Congrats on the launch, wish you success.

this is a category that's going to matter a lot more soon. how do you actually query the models - real API calls on some schedule, or sampling? asking because ChatGPT/Claude/Perplexity answers aren't deterministic, so I'd guess you need a decent sample size per prompt before the citation rate number means anything

Does the tool also recommends what we should do to get more recommendations for them?

Congratulations

Being able to see the exact prompts that led to a mention would be a game changer, even if it's a sampled view. That way we could reverse engineer which question phrasings actually surface our brand versus the ones where we're invisible.

Ran CitedSpy on a small site I work with and was surprised how clear the gap reports were, showing exactly which queries ChatGPT was ignoring us on compared to competitors.

Whats up guys, if you have any questions regarding the product I'll be awake all day responding so please please lmk

Nice launch , qq how does the agent balance token limit optimization when reviewing long multi-page corporate career documents alongside deep user profile sets?

Which data it "reads over the internet" in terms of getting information about the job positions? Also LinkedIn?

Would love to see a calendar integration that blocks out time for interview prep or networking follow ups, since right now the agent handles applications but I still have to manually schedule everything around it.

Would love to see a "quiet mode" where Double only surfaces the top 1-2 matches per week instead of a daily feed. Right now I'm worried about notification fatigue and starting to tune things out, which defeats the purpose of having an agent working for me in the background.

Lot of people use Claude to fix their resumes and apply to jobs. But they still end up doing the search and manually applying. I was wondering how accurate is the AI when it automates this task and does it improve over time ?

Love how the whole experience is just a text conversation — no clunky dashboards or settings maze, just tell it what you want and it handles the rest. That simplicity is hard to nail and they nailed it.

Really interesting concept. How does Double learn from outcomes such as recruiter replies, rejections, and interviews to improve the roles it pursues and the way it represents each user? Can users also see why Double believes a particular opportunity is worth pursuing before it takes action?

Love how the interface keeps everything as a simple text conversation instead of burying the job search behind a dozen menus and filters. Feels way more natural than the usual bloated career platforms out there.

Looks super exciting — congrats to the whole team on the launch! 🚀🚀

Congrats on shipping. The thing I would protect here is the crawler, not the agent.

Luca's answer to Nika is the most interesting line in this thread. Crawling 80k company career pages weekly and skipping the aggregators is a real asset, it is hard to copy, and it is the part nobody in the comments is talking about. Everything else in the pitch, applying for you and growing your network, is something ten other tools are claiming this month.

The risk on the applying side is timing. This is the year hiring teams started drowning in AI applications, and a lot of them are now filtering for exactly that, so volume is turning into a liability rather than an edge. If Double applies to 200 roles for me and 190 are near misses, I have not been represented, I have been mass mailed, and that lands on my name rather than yours.

So the version I would want is fewer and better. Ten roles pulled straight from company career pages, matched properly, with something written that actually sounds like me. That happens to be the story your crawler lets you tell and your competitors cannot.

One small thing on the site. It is worth saying plainly what happens to my CV, and whether anything gets posted or sent as me before I connect an account. For a product that speaks on my behalf, that answer is most of the decision.

Hey everyone! I'm Vance, Double's CTO, and I'll be here all day to answer any questions you have. Also, let me know any features you'd like to see!

Really interesting concept. I’d love to try how Double finds relevant roles and handles outreach.

Really like the focus on quality over mass-applying. A career agent that can tailor roles, applications, and networking from real outcomes feels much closer to what candidates actually need. How do you decide when Double should apply automatically versus ask the user for approval first?

The "just text it" interface is a genuinely distinctive design choice, treating career management like texting a real agent rather than another dashboard to check, which fits your framing of giving everyday people the kind of representation celebrities get. How much upfront context (resume, LinkedIn, past applications) does Double need before it starts acting with good judgment, and can users review outbound messages like network outreach and application copy before they go out, or does it act autonomously by default?

the crawl-80k-career-pages angle is the real moat here, agree with Abdullah on that. curious about the handoff point though - once you're past applying and into something like a take-home assignment or a live coding round, where does Double step back and let the human take over, and does it prep you for that part or just get you to the door?

Upvoted.. Cngrats to launch 🙌 @lucapiekarski putting an active agent in your corner to scout opportunities completely drops the exhausting application anxiety.

auto-applying at scale is the part that worries me a bit - a lot of companies quietly blacklist candidates who spam the same generic application across dozens of roles at their org. does Double tailor each application enough to avoid that, or is it optimizing purely for application volume?

The idea to simplify job search with AI is at least 2 years-old.
Could you take a day or two and think about how would you compete with Jobright AI (launched in 2024 and growing)?

When it applies for me, how much control do I get before something goes out? Recruiters can tell when an application is automated, so I'm curious if I approve each one or if it just runs and I find out later.

the fact that you can just text it like a friend to handle job apps and networking is such a smart way to remove friction. whoever designed that conversational flow really gets how lazy we all are about career stuff.

would love to see a way to track which conversations turned into actual opportunities, like a simple dashboard showing response rates and next steps so i can see whats actually working.

the secret reservation feature is the right instinct for these apps, group gift coordination falls apart fast without it. curious if there's an iOS app planned too, or if the shareable web link is meant to make that unnecessary

the secret reservations bit is the actual hard part of these apps to get right, my family's tried a couple wishlist tools before and they always leak who reserved what if two people open the list at the same time. how are you handling that race condition on the backend?

This is a very practical idea for families. Wishlists are easy to make, but coordinating gifts across parents, grandparents, siblings, and friends can get messy fast.

I like that Farha works with any store and can be shared by link without requiring everyone to sign up. That makes it much easier for extended family members to actually use it.

Quick question: how do secret reservations work in practice? Can gift givers see what has already been reserved while keeping the surprise hidden from the person receiving the gift?

Most people only think about wishlists a few times a year, so retention will be a bit difficult. However, I like the group gifting feature.

Love that sharing a wishlist doesn't require the recipient to sign up for anything. That friction-free link sharing is such a thoughtful UX choice, especially for older relatives who just want to see what someone wants.

A friend mentioned this the other day and I gave it a try, the secret reservation thing is such a smart touch. One thing that would take it further for our family is a built-in budget tracker per wishlist, so anyone contributing to group gifts can see how close we are to the goal without pinging the organizer. Would save a lot of back and forth in our group chat.

the secret reservation feature is honestly so smart, finally no more awkward double gifts at birthdays. set up a quick list for my sister's birthday and sharing it without her needing to sign up is super smooth.

Finally tried this with my sister's birthday coming up. Adding stuff from random stores without having to make accounts was honestly the best part, and my brother saw the same Bose headphones and chipped in without telling her. Pretty handy.

Really clean experience adding items from random stores and sharing the link with my sister worked without her needing to sign up. The group gifting idea is surprisingly practical for my parents' birthdays.

A small thing that would make sharing smoother is letting us set an expiration date on a wishlist link. Family events come and go, and it would be nice to have links auto-close after the birthday or holiday is over so old lists stay private without me having to remember to delete them.

Adding an in-app birthday or event reminder that pings family members a week before the date would be super helpful. Right now I still rely on Google Calendar to remember my niece's birthday, and it's easy to miss. If Farha could nudge everyone in the shared group automatically, it would become the one place my family actually checks for gift planning.

A nice simple gift-coordination app. One feature that would really help during birthdays and holidays would be letting us schedule a "reveal date" for secret reservations so they automatically unlock the moment a celebration arrives, no manual nudge needed.

the secret reservation feature is such a smart touch, finally a way to avoid the awkward "i got the same gift" moment at family birthdays.

Love that the share link works without forcing recipients to sign up, such a respectful touch for family members who just want to peek at a list. The secret reservation feature is clever too.

Love that shared links need no signup. Makes it so easy to drop a wishlist in the family group chat without making everyone create accounts just to peek.

Congrats on the launch. One thing I'd love to see is a browser extension that lets me add items to a wishlist directly from any shopping site with one click, instead of pasting links manually every time. Would make building lists for birthdays and holidays way faster for busy parents like me.

The secret reservation idea is clever for avoiding duplicates. One thing I'd love to see is a built-in birthday or occasion reminder that nudges family members a week before so we actually check the wishlists and buy on time. Would make the whole coordination flow feel less last-minute.

The no-signup sharing is a lifesaver for sending my mom a quick birthday list without making her create an account. Love that the secret reservation feature prevents my sister from buying the same thing twice again.

The secret reservation feature sounds really clever for avoiding duplicate gifts. One thing I'd love to see is a price tracking option so the wishlist owner gets notified when something goes on sale, and contributors can see the price history before chipping in for group gifts.

Browser extension so I can add items to a family wishlist without leaving the store page would save a lot of time. Right now copying links over feels like extra steps that could be skipped.

Finally tried Farha with my sister's birthday and the secret reservation feature was a lifesaver—no more awkward duplicate gifts from family. Love that sharing the wishlist didn't make everyone create accounts.

Finally tried Farha with my family and the secret reservation feature actually works smoothly—no awkward duplicate gifts this year. The no signup link share is a nice touch.

Love that the link sharing works without requiring a signup on the other end, that removes the biggest friction for non-techy relatives and actually makes the whole gift coordination thing usable in real families.

an agent that writes and publishes articles daily to match search intent sounds exactly like the pattern Google's helpful content update was built to catch. has anyone running this for a few months actually seen rankings hold up, or does the traffic show up short term and then get filtered out once the volume looks automated?

the "publishing them for you" part is the piece I'd want to understand better - does the agent auto-publish straight to the live site, or queue drafts for a quick human approval first? auto-publishing unreviewed AI content directly onto a domain feels like it could cut either way for SEO depending on how tight the review loop is

finally gave the AI SEO agent a real shot and it actually published two articles overnight that match my site's tone pretty well. the cookieless tracking is a nice bonus too

the cookieless setup without that annoying consent banner is such a smart move honestly, makes the whole experience feel way cleaner from the jump

Really like how the AI agent actually publishes the articles for you instead of just handing you a draft, that's a level of execution most tools skip. Also the cookieless setup is genuinely refreshing, no consent banner feels almost luxurious these days.

Clean dashboard and the cookieless angle is genuinely refreshing, especially with consent banners becoming a pain across EU traffic. Curious to see how the AI-written articles hold up over time, but the keyword research alone feels worth a trial.

The cookieless angle is a huge plus and the built-in SEO agent sounds really useful. One thing that would make this a no-brainer for me is a way to compare traffic and conversion data side by side with my previous GA setup, so I can actually see whether the switch is paying off and by how much.

Finally tried a privacy-friendly analytics tool that actually loads fast and the built-in SEO agent already drafted two decent articles overnight. Surprised how clean the traffic source breakdown looks without the usual cookie banner headache.

The cookieless setup was refreshing, no banner hassle at all. The AI agent actually published a decent first article overnight without me babysitting it.

The cookieless setup without a consent banner is such a relief compared to juggling GA4 configs. Really nice execution on making the SEO agent feel like it actually does the research and writing instead of just spitting out generic drafts.

Tried it out for a day and the keyword research actually felt tailored to my niche instead of generic suggestions. Liked that the real-time view doesn't need a consent banner, saves me from those annoying popup headaches.

Ran the install in under a minute and it immediately flagged two skills I'd been using for months as sketchy. Wild that something this thorough is free and on-device.

finally something that actually scans the skills i already installed instead of just trusting them. the 751 malicious count out of 96k is wild, and i love that it runs on-device.

ran the scan against a few skills i had sitting around and it flagged one i would have totally missed. love that it just runs locally without sending anything off-box.

ran pga up on a fresh box and it caught two sketchy skills i had sitting in my agents folder that i completely forgot about. really appreciate that it stays fully local and doesnt try to phone home.

Love that this finally exists, scanning 96k skills and catching 751 malicious ones is no small feat. One thing I'd find super useful though: a watch mode that monitors new skills published to common registries in real time and pings me when one matching my installed list gets flagged retroactively. Would close the gap between scan day and the next malware drop.

honestly the runtime hijack blocking is the part that sold me. one thing though — could you add a simple diff view when panguard flags a skill as suspicious? like show which ATR rule tripped and what line in the skill triggered it. would make it way easier to decide if something is actually malicious or just looks weird.

Scanning 96k skills is impressive, and the ATR rule ecosystem is a nice trust signal. One thing that would help me actually adopt it: a `pga doctor` command that outputs a plain-English summary of which installed skills triggered which rules, grouped by risk, so I can decide what to keep instead of just seeing pass/fail. Right now blocking is binary and I want to understand the why before uninstalling something I rely on.

solo builder, self-taught, four months in, and already merged into Microsoft/Cisco/MISP/OWASP tooling is a wild pace. one question on the runtime watcher piece specifically, does watching the agent at runtime add any noticeable latency to its actual work, or is it lightweight enough to just leave running in the background all the time

751 out of 96k is a much higher malicious rate than I expected, that's a genuinely useful number to have public. for the runtime hijack blocking specifically, is that based on known attack signatures or are you also catching novel prompt-injection-style hijacks that don't match an existing rule yet?

love that you put real numbers out there instead of hand-wavy claims, the 751 out of 96096 figure is the kind of evidence that makes this feel like a serious tool rather than a pitch. shipping it open source with the ATR rules already folded into MISP and OWASP is genuinely impressive execution.

ran the one-liner and it flagged two sketchy skills i forgot were even installed, kind of unsettling but honestly glad i know now. the on-device scan is a nice touch.

ran pga up against my usual skill folder and it flagged two MCP wrappers i totally forgot about, the on-device scan is genuinely nice. honestly appreciate that it's MIT and just one command away.

ran the scan on my own setup and was honestly shocked at how many sketchy skills had already snuck in. the one command setup made it painless.

Ran it locally for a few days and liked that my own machine handled the comments instead of pinging some random server. The LinkedIn comment suggestions actually read like something I might type, which surprised me more than it should have.

AI-generated comments at scale is the part that gives me pause here, LinkedIn's spam detection has gotten a lot better at flagging generic-sounding replies, and Instagram is even stricter about automated engagement. running locally instead of through their API probably helps with rate limits, but does it actually reduce the chance of a comment reading as bot-written to another human, or is that still on the user to catch before it posts?

Adding sentiment analysis to the AI comments would be a nice touch, so it can avoid sounding too salesy or off-key depending on the original post's tone. Would help the engagement feel more genuine and reduce the chance of awkward replies.

ran it for a couple days on desktop and the ai comments actually feel contextually relevant, not the usual generic “great post” spam. scheduling linkedin posts side by side with instagram has saved me a lot of tab hopping.

Would love a built-in lead scoring layer that flags commenters who match your ICP, so the AI can prioritize replies to high-fit accounts instead of treating every interaction equally.

Running it locally on my desktop is a nice touch, feels way safer than handing over my login to some cloud tool. The AI comments actually read pretty natural, not the usual spammy stuff I've seen elsewhere.

Runs locally which is a nice plus, comments actually sound like a real person rather than the usual AI fluff. Surprised how well it picks up on context.

Love that it runs locally on desktop instead of pushing everything through a browser tab, smart call for anyone nervous about account safety. The scheduled post + AI comment combo looks tight too.

Love that this runs locally instead of being yet another cloud service farming engagement data. That's a real differentiator for anyone worried about LinkedIn flagging their account or getting burned by API changes.

Would love to see a sentiment filter that flags low-quality or spammy posts before the AI comments on them, so the tool avoids accidentally boosting bad content on your behalf.

Love that it runs locally on the desktop instead of pushing everything through a cloud. That choice alone shows the team actually thought about trust and consistency for B2B users, not just flashy AI features.

the trust angle is smart, especially the story about matching an actual mechanic quote. one question - repair labor costs vary a ton by region and even by country, does the cost estimate localize to where you are, or is it working off a more generic baseline right now?

Snapped a pic of my squeaky belt and it nailed the issue in seconds with a cost range that actually matched what my mechanic quoted last month. Honestly didn't expect the recommendations to be that on point.

honestly really cool idea, would save me so many trips to the mechanic. one thing tho, would be awesome if i could upload a short video or a few photos from different angles since sometimes one pic doesn't really show the full issue. that would probably make the diagnosis way more accurate

Snapped a pic of my squeaky brake caliper and got a pretty accurate guess within seconds, plus a parts list that actually matched what my mechanic quoted last month. Honestly impressed for the price.

love that you can just snap a pic of a part instead of digging through forums trying to describe the noise. the instant cost estimate alongside the diagnosis is a really smart touch, removes the whole back-and-forth with the shop.

snapped a pic of my squeaky brake and got a solid answer in like 10 seconds, way more useful than scrolling through random forum posts honestly

Snapped a pic of my weird rattling exhaust and it actually nailed the issue plus gave me a fair price range before I called the shop. Wild that it costs less than my morning latte.

Snapped a pic of my sputtering alternator and got a clear diagnosis plus a price range that actually matched what my mechanic quoted later. Solid little tool for anyone who hates getting blindsided at the shop.

honestly this looks super useful for anyone who doesn't know much about cars. one thing I'd love to see is a "before you go to the mechanic" checklist or some kind of prep guide so you know what to ask or verify when you take it in. would save a lot of people from getting upsold on stuff they don't actually need

snapped a pic of my weird rattling thing under the hood and it actually nailed what was wrong, plus gave a realistic price range. kind of wish i had this before my last shop visit honestly

Snapped a pic of my weird rattling exhaust and it nailed the issue in seconds, even threw in a rough cost range that matched what my mechanic quoted. Honestly didn't expect it to be that spot on for the price.

Snapped a pic of my squeaky belt and it nailed the issue right away, even gave me a rough price range. Way easier than scrolling through forums trying to figure out what's wrong.

One thing that would really help is adding a way to track repair history for each car over time. So I could upload the same brake pad photo in six months and see how wear progressed, or check what fixes were recommended last visit. Would make it way more useful as an ongoing tool instead of a one-time check.

Hello Product Hunt 👋

Modern work doesn’t happen in one app anymore.

A prompt in ChatGPT.
A design in Figma.
A file in Finder.
A link in Safari.

The hard part isn’t finding information.

It’s keeping the right pieces close as you move between apps.

That’s why I built Tansei.
It’s a simple shelf for your Mac.

It automatically keeps the text, links, images, files, code, colors, and screenshots you copy, so they’re always within reach. Drag them into any app, paste them with your keyboard, or come back to them later.

Everything stays on your Mac. No account. No cloud.

I’d love to hear how you work:

• What’s always open on your Mac?

• Which apps do you switch between the most?

• What’s one thing you copy over and over every day?
  

Thanks for taking a look — I'll be here all day.

since it auto-captures everything you copy and keeps it all local, does the shelf have any retention or auto-cleanup, or does it just grow forever until you manually clear it? curious how that's handled once you've been using it for months

honestly the pinned-shelf idea looks super handy, but one thing i'd love is a quick keyboard shortcut to call up the shelf and search through pinned items by content, like the actual text from chatgpt or the hex value of a color, instead of just the app it came from

One thing I'd love is a quick keyboard shortcut to pop the shelf open from anywhere, even when it's hidden, so I can grab a clip without losing my mouse position.

love how local and no-account this is, finally. one thing that would honestly make it stick for me is keyboard shortcuts to pin and recall items without reaching for the mouse, like a quick hotkey to grab whatever's currently on the clipboard into the shelf

Love how lightweight this feels and the local-only approach is a huge plus. One thing I'd find really useful: let me group pinned items into named stacks so I can separate work-in-progress from quick-reference snippets, and collapse the whole shelf with one keystroke when I want a clean desktop.

Love that it stays tucked at the edge instead of interrupting your flow. Most shelf apps feel like another dock, but the pinned snippet idea (especially grabbing a line of chat output) is genuinely clever.

the way it just lives at the edge of the screen and gets out of the way is honestly so well done, feels like it was thought through by people who actually use macs daily.

Love that it runs locally with no account, that's exactly what I'd want for something this personal. One thing I'd kill for: a quick hotkey to pin from anywhere, not just drag and drop. Like a global shortcut that grabs whatever's currently in my clipboard without me hunting for the app window first. Would make the whole "without losing the thread" promise even more true.

Pinned a color and some text from ChatGPT to test it out, and dragging them back into other apps felt genuinely smooth. Love that everything stays local, no account fuss.

Pinned a color hex and a ChatGPT snippet earlier and honestly forgot how handy that is when bouncing between apps. Nice that it's all local too, kind of rare these days.

Love the privacy-first approach and the idea of pinning stuff at the screen edge. One thing I'd really want is a keyboard shortcut to instantly summon the shelf and search across pinned items, instead of hunting with the mouse when I'm mid-flow. That'd make it feel even more invisible until I need it.

The deterministic 7MB rule engine over the Accessibility API is the right tradeoff here — no model means nothing to phone home, which is the whole point for a leak-prevention tool. Since Cursor and Copilot-in-VSCode are Electron, does the composer read stay reliable there, or does the Accessibility tree get flaky compared to a native field like Claude Code in the terminal? And can I add rules for our own internal token prefixes locally, or does expanding past the 47 built-ins need an app update?

One thing I'd love to see is a simple CLI flag so I can pipe clipboard content through HeimWall's rule set from my terminal without opening the app. That way quick checks during code review stay in my flow.

The on-device approach is a big deal for trust. One thing that would help me roll this out to my team is a dry-run mode that lets me see which of my recent prompts would have been flagged, so I can fix habits without the awkward retroactive alerts.

would love to see a quick test mode where i can paste in a sample prompt and see exactly which rules flagged it, basically a way to tune the 47 rules for my own workflows without needing to trigger real leaks in my actual coding sessions

Really like the signal over content approach. One thing that would help me actually roll this out is a one-click allowlist for the specific secret values I know are safe but get flagged constantly, like local Postgres URLs or my dev API keys. Right now I'd imagine false positives get noisy fast. Persistent per-project overrides with maybe an audit log would make this feel less like babysitting and more like a real safety net.

Love that it's fully on-device, finally something that doesn't sell my clipboard to a cloud. One thing I'd want soon though - a way to whitelist my own throwaway test secrets so I'm not hitting the red on every dummy API key I paste in for debugging.

Have you thought about adding a quick toggle in the menu bar to temporarily disable blocking when I intentionally need to paste a real API key during local testing. A 15 minute pause button would save a lot of friction compared to fully uninstalling and reinstalling every time.

Ran it for an afternoon while pushing code through Cursor and it actually flagged a stray API key I had no business pasting in. The "fully on-device" bit sold me, no setup, no signup, just a quiet macOS app doing its job in the background.

the on-device approach is genuinely refreshing, especially with so many tools phoning home these days. one thing that would make this a no-brainer for me would be git hook integration so it can scan staged diffs before they ever reach an AI tool in the first place, catching leaks at the source rather than only at the prompt boundary.

love that this stays fully on-device, that was basically the deciding factor for me. one thing though, would be great if you could add a quick toggle to whitelist specific projects or repos, like a dotfile in the project root, so it doesn't flag test fixtures or seed data as real leaks. right now i imagine anyone with mock credentials in their codebase will get noise.

honestly this looks really useful for anyone paranoid about pasting real keys into copilot by accident. one thing i'd love is a quick "why was this flagged" popover when a rule triggers, basically a one-liner explaining which pattern matched so i can learn what to scrub next time. would make the whole thing way less mysterious

the DevGPT stat sold it, three live keys and 49 emails in one week of public snapshots is a genuinely alarming stat. since detection is rule-based rather than a model, how do you handle secrets with no recognizable format, like an internal API key using a company-specific naming scheme that just looks like a random string? that seems like the hardest case for a pattern-matching approach

Have you thought about adding a way to whitelist specific channels or topics so they never get filtered out, even if flagged as AI? Would be nice to keep learning channels I trust without having to re-pin them every time the filter runs.

honestly the icon is super clean and the whole "kill the slop" framing feels way more honest than most feed-cleaner tools out there, like you're not beating around the bush about what you're targeting

Took it for a spin on my usual mess of a feed and it actually pushed AI-generated junk out of my recommendations within a day. Nice that it runs locally without sending my watch history anywhere.

Love the concept, especially filtering out AI generated stuff. One thing that would make this a must install for me is letting me whitelist certain channels or keywords so my favorite creators never get hidden by mistake, maybe with a simple toggle in the popup.

Would love to see a toggle that lets me hide channels I've already watched from yesterday's feed since I keep scrolling past them anyway. Maybe also a quick daily recap showing how much AI slop got filtered out so I know the extension is pulling its weight.

The way it lets you filter by content type rather than just hiding channels is a thoughtful UX move, feels way less aggressive than other tools I've tried.

the false positive side feels like the harder problem here honestly. plenty of human creators now use AI somewhere in their pipeline, b-roll, thumbnails, voice cleanup, without the video actually being AI-generated in the sense you mean. does the detection separate AI-assisted from AI-generated, or is that still a fuzzy line for you too

This is a very timely problem. YouTube feeds are getting harder to navigate as AI-generated videos, synthetic voiceovers, spam, and bot comments become more common.

I like that KillAI focuses on giving viewers more control rather than just complaining about the trend. The privacy-first angle also matters a lot for a browser extension.

Quick question: how does KillAI handle false positives? For example, if a human creator uses some AI-assisted editing or voice cleanup, how do you avoid hiding content that is still genuinely human-made?

The way you scoped it down to just YouTube instead of trying to boil the ocean feels really intentional. Liking the focus on surfacing actual creators and discussions over everything else.

Would love a toggle to whitelist specific channels I trust, so I don't have to keep manually approving creators I already know are real. That way the filter keeps working but I don't miss uploads from my favorite smaller creators.

It would be really useful if you could whitelist specific channels or topics that you actually want to keep seeing even after a cleanup pass, so the extension preserves your favorite creators automatically instead of hiding them too.

the suggested sensitive area detection is the part I'm most curious about. how good is it at catching things like emails or api keys sitting in a screenshot, or is it mostly flagging obvious stuff like faces and the text-based secrets are still on the user to catch manually

Cleaned up a product landing page in under a minute and the auto-redact suggestion caught an email i hadnt even noticed. Nice that everything stays local.

Snapped a screenshot of my portfolio and the curated backgrounds actually made it look launch-ready in seconds. Local processing is a nice touch for sensitive client work, though the ₹1,899 unlock feels a bit steep after just 10 exports.

Captured a pricing table from a client's site and the drop shadow plus subtle border made it look like a proper launch asset in seconds. Local processing is a nice touch for anyone worried about pasting sensitive dashboards into yet another cloud tool.

Captured a pricing page in one click and the shadow and browser frame options made it look honestly ready to post, no Photoshop needed. The local processing is a nice touch too.

The local processing is a big plus. One thing that would make it way more useful for me is keyboard shortcuts for the capture modes and style presets, so I can flip through backgrounds and shadows without breaking flow. Right now I'm mouse-hopping a lot between the dropdowns.

A nice twist would be letting users save their own style presets alongside the curated ones, so a brand color combo or shadow setup doesn't have to be redone every capture. That plus a simple way to reuse the same redactions across exports would save a lot of clicks.

Being able to save my own preset combinations of styles and backgrounds would save a ton of time when capturing similar elements across pages.

Love that the redaction step makes you confirm sensitive areas before exporting. The curated style presets feel considered too, not just generic filters slapped on a screenshot tool.

honestly the local processing angle is a really thoughtful call here, love that nothing leaves your machine. and the curated styles plus those ready-made export sizes basically skip the whole back-and-forth of resizing for different platforms.

Captured a section of my product page and the preset styles looked way more polished than my usual screenshots, the local processing was a nice surprise too.

Captured a pricing section from a client site and the drop shadow plus rounded corners made it look like a proper mockup in seconds. The local processing is a nice touch for anyone nervous about pasting URLs into random tools.

Would love to see a built-in secondary marketplace view that shows live bid/ask spreads for each tokenized round, would make liquidity much more tangible for newcomers comparing opportunities at a glance.

One thing that would honestly help is showing a fee breakdown right on the investment screen, like gas costs, platform fees, and any spread on tokenized equity. Right now it's kind of unclear what you're actually paying for a trade, and that transparency would probably build a lot more trust with folks new to this kind of investing.

Congrats on the launch. One feature I'd love to see is a built-in secondary marketplace filter so I can sort tokenized startups by sector, stage, or minimum check size. Right now browsing raw offerings feels like a lot of digging, and that would make comparing opportunities way easier.

The AST-CodeSplitter plus LangGraph retrieval is what makes this more interesting than a generic summarize-my-repo wrapper — parsing real structure instead of flattening everything into embeddings should hold up better on an actual monorepo. When I point it at a private repo, does it clone the whole thing to your backend to run the split, or can I scope it to a subtree? And can I export the generated docs and Mermaid diagrams as markdown/.mmd to commit back into the repo, or do they only live in the persistent history inside your app?

Would love to see some kind of secondary market price history chart for each tokenized startup so we can track how valuations move over time. Right now it's a bit hard to gauge momentum or spot trends before buying in, basically just guessing. A simple sparkline or longer term chart on each listing would make the whole experience feel way more transparent and help folks make more informed decisions.

Investing in startups feels way less intimidating when there's actual liquidity built in. Curious to see how the tokenized equity trading works in practice once a few real deals go live.

the tokenized equity concept is pretty cool, especially the 24/7 liquidity angle for VCs. landing page loaded fine but honestly wished there was more detail on the actual trading flow before signing up.

The tokenized equity angle is genuinely interesting, especially the 24/7 liquidity part since traditional VC has such long lockups. Wish the site had more detail on how secondary market trading actually works in practice.

It would be great if you added a simulated portfolio feature where users can test tokenized equity trades with fake money before committing real funds, especially helpful for newcomers trying to understand how startup valuations fluctuate over time.

Would love to see a secondary market order book view with depth charts directly in the dashboard. Right now it's hard to gauge real liquidity beyond the 24/7 claim, and visualizing bid/ask spreads would make the tokenized equity trading feel a lot more transparent before committing capital.

the tokenized equity angle on Polymesh is honestly a smart call, feels like they actually thought through the compliance side instead of just slapping "decentralized" on everything

love the idea of letting VCs actually exit without waiting for an acquisition, the polymesh backing adds real credibility. curious how the secondary market for these tokens actually holds up when a startup is still private.

Honestly love how clean the layout is here, no clutter just the essentials staring back at you. The fact that it tracks goals alongside income and expenses in one place is a really thoughtful touch, you know, feels like the team actually thought about how people budget day to day.

Love how clean and uncluttered the layout is. The single-page view for income, expenses, and goals feels like it actually respects how people think about their money.

Love how clean the layout is, especially the way income and expenses sit side by side without feeling cluttered. The goal tracking section is a smart touch for keeping motivation visible.

Love how clean the layout is here, no clutter getting in the way of actually seeing where your money goes each month. The simple category breakdown makes it feel approachable even if budgeting feels intimidating at first.

honestly love how clean the layout is, every section is right where you'd expect it without any clutter. the way the categories are organized makes it feel actually usable instead of overwhelming, which is kind of rare for budget tools.

Tried it for a week and the income vs expense view actually made me notice where my money was leaking. Simple enough that I stuck with it past day one, which says something.

Would love to see a built-in bill reminder or notification system so I get a heads up before due dates hit. Right now I have to cross-check my budget against my calendar, which kind of defeats the purpose of having everything in one place.

The single page layout is genuinely smart, keeps everything in view without making you dig through tabs. Nice that savings and goals sit right next to expenses instead of buried in some submenu.

Really liked how clean the income vs expenses view is, made it easy to spot where my money was actually going each week.

Been using something similar and what really helped me was a visual progress bar for each savings goal, so I can see at a glance how close I am without doing the math myself. Could you add that to show goal percentage completion right next to the target amounts?

The clean layout makes it easy to plug in expenses without feeling overwhelmed, and I like how the savings goal sits right next to the spending breakdown so I can see the tradeoffs.

Finally something that doesn't overwhelm me with categories. The monthly view is clean and adding expenses takes seconds, which means i'll actually keep using it.

A simple budget planner like this is genuinely useful, but I'd love to see a category for irregular or one-off expenses so I can plan ahead for things like car repairs or yearly subscriptions without throwing off the monthly totals.

honestly looks solid for keeping tabs on monthly stuff, but one thing that would be super useful is an overdraft or low-balance warning when you get close to your spending limit before the month ends. helps you course correct instead of finding out after the damage is done.

Love how clean the layout is, separating income, expenses, and savings into their own clear sections makes it feel way less overwhelming than other planners I've tried. The monthly focus is a smart call too.

finally tried this after ignoring my finances for way too long, super clean layout and actually makes inputting expenses painless. wish i had something like this months ago

finally found something that does not overwhelm me with charts i do not need, just a clean grid for income and expenses that i can fill in each month.

Made a few candle labels in about ten minutes with just my home printer, and the designs actually looked clean on the sticker sheets. Wish I had found this sooner instead of wrestling with Word templates.

Finally a label maker that doesn't need special equipment, just used it for my soap jars and the templates were actually easy to tweak. Might stick with this one.

honestly this is kind of a game changer for my candle side hustle, finally I can print decent labels on my regular printer without dealing with that expensive thermal printer nonsense

Finally something that saves me from buying yet another gadget. Dragged in my candle logo and it just printed clean on regular sticker paper, no fiddling with margins.

Love that this avoids the cost of a label printer - super useful for small batches. One thing that would make it way better for me: a way to save label templates with my own dimensions so I don't have to re-enter the same measurements every time. Would save a lot of clicking across batches.

The fact that it works with any sticker sheet you already have is such a smart move, honestly most small makers have half a pack lying around already. Really nice that you don't need to drop cash on a fancy label printer just to get started.

honestly the no-special-printer thing is kind of a game changer for my candle side hustle, used it on regular sticker sheets and it looked pretty legit right out of my home printer.

Been wanting something like this for my soap business. One idea would be to let users save their label designs to an account so they can come back later and tweak them, instead of redoing everything from scratch each time. Would save a ton of hassle for folks doing repeat batches.

Would love a simple inventory tracker built in so I can see how many of each labeled batch I have left. Right now I design the labels here but have to bounce to a spreadsheet to keep stock straight.

Honestly this is kind of a game changer for my candle side hustle, no more begging my sister to use her label printer. The templates are simple and it just works with my regular inkjet, took me maybe ten minutes to design a full sheet.

This is super useful for my little soap side hustle, no more buying expensive label paper. One thing that would help a ton is a built in QR code generator right inside the editor so I can link to ingredient lists or my shop without opening another tool.