I build software mostly on my own, so developer tools are the launches I read most closely.
In the first 18 days of October, Junior picked up 38 of them from Product Hunt's daily lists.
Three things stood out. Agents are getting their own homes (a browser, an app, a terminal). Agents want to take your on-call shift. And a surprising number of developer tools are about video.
Here are the ten with the most upvotes, what each one does, and who it's for.
1. Rill Browser
515 upvotes, #1 on October 6
A browser built around Claude Code and Codex.
You browse normally. When you see something worth doing, a bug report, a docs page, a design, you press ⌘E and it becomes a task for your agent. Rill finds the right project, passes the context along, and the agent works while you keep browsing.
It also sorts your tabs into themes and turns your history into days you can search. Free, on the agent plan you already pay for.
Who it's for: developers who already run coding agents and lose time copying context from the browser into the terminal.
The most upvoted developer tool of the month so far, and I get it. It's solving a problem I have every day.
2. Inferock
401 upvotes, #2 on October 14 · inferock.ai
Serverless LLM inference, pay per token, with one twist: you get credits back on eligible failed calls.
You can also bring your own OpenAI, Claude or Gemini keys and see where calls fail and what they cost. New accounts get signup credits to try every model.
Who it's for: teams paying for a lot of LLM calls and not sure how many of them are failing.
3. Onepin
383 upvotes, #1 on October 10 · onepin.com
AI voices sound human now. But they still guess names from the spelling.
Onepin sits on top of the voices you already use (ElevenLabs, OpenAI, Google and 30+ more). It spells out prices and dates, checks names against a 4M-word pronunciation dictionary, scores every line, and fixes the wrong word without re-rendering the whole take. Free to start.
Who it's for: teams shipping voiceover in several languages: product videos, dubbing, courses.
4. Claude Haiku 5.5
344 upvotes, #3 on October 8
Anthropic's fastest and most capable small model, built for high-volume, cost-sensitive work: summarization, classification, coding subagents, customer support and browser use.
Who it's for: anyone running the same small task thousands of times a day, where the big model is too slow or too expensive.
5. Polylane (two launches)
197 upvotes on October 1, and 116 on October 15 · polylane.com
Polylane launched twice this month, and both made the list.
The first launch is an agent that handles incidents. It connects your code, infra and observability data, investigates every incident, and opens a pull request with the fix. If code can't fix it, you get the root cause and a recommendation. Their line: "Nobody should be on call."
The second, Impact Analysis, works before the incident. It reviews every pull request against your live infrastructure (config, capacity, dependencies, telemetry) and tells you, with evidence, whether it's safe to merge.
Who it's for: small engineering teams where "on call" means the founder's phone.
Launching twice in one month and placing both times is a good reminder: one launch isn't the only shot you get.
6. Video Editor in Manus 2.0
112 upvotes, #13 on October 7
Manus added a video editor. The team says that after a month of daily use, people made launch films, game trailers, news explainers and beat-synced animations with it, and edited travel footage.
Who it's for: developers and founders who need a launch video and don't edit video.
7. iphone-use
109 upvotes, #11 on October 6
Gives an AI agent a real iPhone. The screen comes back as text, and the agent can tap, type and scroll, even in apps that have no API.
Two details I liked. Every action gets an honest answer: applied, not sent, or unknown. And a task done once becomes a flow that replays without calling a model. Payment apps that block screenshots come back as a wireframe.
It ships with an HTTP API and a 21-tool MCP server for Claude Code. Open source, MIT.
Who it's for: anyone testing iOS apps, or automating something that only exists on a phone.
8. Refs
108 upvotes, #8 on October 9
A library of 3,000+ films, each measured cut by cut and written up as a 15-part blueprint with a prompt. Add it to Claude Code, Codex or Cursor with one command, and your agent plans your launch film from real examples instead of guessing.
Who it's for: developers using an agent to make a launch video, which is apparently a lot of developers now.
9. Genchi
105 upvotes, #11 on October 13 · genchi.com
Predicts which projects are about to miss their deadline, by asking the people doing the work.
Team members vote with one anonymous click in Slack. Leaders get a combined score and trend for every project, in one view. It works with Jira and Asana, and you can check your portfolio from inside Claude, ChatGPT or Grok.
Who it's for: engineering leaders with more projects than they can follow closely.
The only tool in the top ten about people, not code. Very nice to see.
10. 3AM Super Intelligence
100 upvotes, #7 on October 18
An AI on-call engineer "for systems you can't rewrite". It finds the root cause, rehearses the fix, and resolves the incident with one tap in Slack, all from inside your network.
Who it's for: teams running older systems that can't send their data to an outside service.
Three more worth a look
- Parallax (99 upvotes) runs Claude Code, Codex and 40+ other coding agents in one app, with 240+ models on your own subscriptions. When one subscription runs out, your work moves to the next.
- Buddy Drop (98 upvotes) puts a site online in seconds. Drop a folder, ZIP or files, pick a free domain or connect your own. No signup.
- Dropstone CLI (90 upvotes) is a terminal coding agent that remembers your corrections across sessions, and turns a GitHub issue comment into a pull request. It's on every plan, including free.
What October is telling us
Agents are getting a home.
Rill gives them a browser. Parallax gives them an app. Dropstone gives them a terminal. iphone-use gives them a phone. A year ago the question was which model to use. Now it's where the agent lives and how it gets context from you.
On-call is the next job agents want.
Polylane, Polylane again, and 3AM. Three launches in the top ten that investigate production problems and propose fixes. They all keep a human in the loop: a pull request to review, a tap in Slack. Nobody is selling "let the AI push to production" yet.
Developers are making videos.
Onepin, Refs and the Manus video editor are all here. Every launch needs a video now, and developers would rather build a tool than learn an editor. Could be a me problem, but that sounds right.
If you sell to developer tool teams
Most of these are small teams who just put their work in public. They're watching signups, reading every comment, and figuring out what to fix and what to buy.
That's why I built Junior (not affiliated with Product Hunt). Each launch comes with the makers behind it, so you can reach the actual person who built it.
The full list
This is the top ten of 38. The full ranked list of October's developer tool launches, with makers, is here: The Best Developer Tools of October 2026. It updates every morning until the month closes.
Which one are you trying first?
Upvotes and ranks as of October 18, 2026. Product details come from each team's own launch description.