922 projects. One searchable page.
Every month, 900+ developers describe what they're building on Ask HN "What are you working on?". HN Radar structures all of it into a searchable, filterable, exportable radar.
Free web version. No spam — one email when the next issue drops.
Why Developer Project Data Matters Right Now
Ask HN "What are you working on?" is one of the highest-signal threads on the internet. The September 2026 edition alone pulled in 922 comments and 292 upvotes, and it is not a one-off — it runs every single month. That means hundreds of developers are voluntarily telling the world what they are building, what stack they chose, and where they are stuck. This is not passive complaining or secondhand opinion. It is first-person disclosure from people who have already sunk real time into a project, which makes it the most honest demand signal available anywhere.
The problem is that nobody can actually read it. A 922-comment thread is a wall of unstructured text. The projects are buried between tangents, the tech-stack choices are scattered across replies, and the recurring pain points — billing integration, auth, App Store review — never surface as patterns. Existing tools do not help. HN's native page shows the raw comment tree. Aggregators like hnrss and hckrnews work at the post level, not the comment level. Newsletters like TLDR and Pointer summarize the news, not the projects. Crunchbase covers funded companies. Product Hunt covers launched products. None of them tell you who is writing their first line of code right now.
That gap is why this moment matters. The data source is stable, the format is predictable, and the audience is already reading the thread — just painfully. Early investors want deal flow months before Product Hunt. Technical leads want to know which stacks are actually being chosen, not which ones trend on Twitter. Indie developers want to avoid rebuilding something that already exists. All of them are currently doing this by hand, one comment at a time. HN Radar turns that manual grind into a structured dataset that updates every month, so the signal compounds instead of resetting.
How It Works
We pull the full comment tree
Each month's thread is fetched directly from the HN Algolia API, capturing every project self-description, reply, and link without scraping HTML or losing nested context.
We classify and structure it
Comments are parsed into project cards with a name, one-line description, tech-stack tags, links, and category — then aggregated into a stack co-occurrence matrix and a pain-point cluster report.
You search, filter, and export
The result is a single static page with fuzzy search, tag filters, and one-click CSV or JSON export, so you can slice the data however your workflow needs it.
What You Get
Searchable project cards
Every self-described project becomes a card with a name, description, stack tags, and links. Fuzzy search and tag filters let you find competitors, collaborators, or adjacent ideas in seconds instead of scrolling 922 comments.
Tech-stack co-occurrence
See which combinations are actually being chosen: Rust plus SQLite appearing fourteen times, Next.js plus Supabase nine times. This is real selection data from builders, not trend noise from social feeds.
Pain-point clustering
Sentences around "struggling", "stuck", and "hard part" are clustered into the top recurring blockers — billing integration, auth, App Store review — so you can see where developers actually need help.