We just shipped the new @superlogYC. And it runs on your subscription. It does your routine engineering work: triaging alerts, fixing production errors, answering support questions, scanning for reliability issues. Any model. Any harness. Runs on your AI subscription, so you pay nothing for inference. superlog.sh

we just launched Responder on @ProductHunt! (link's in the first comment 👇) quick story: every observability tool is great at exactly one thing, telling you something broke. then it hands you a wall of logs and wishes you luck. we got tired of that. so we built Responder. now live on Product Hunt. an upvote and a comment would be super useful!

Hey everyone, we just shipped Superlog Responder. Our bug-fixing framework, now on top of Sentry/Datadog. TL;DR: Some of you wanted our >90% PR-acceptance framework but not another observability tool to install. We got it. Responder plugs into your existing Sentry/Datadog Slack channel, triages every alert, and replies with root cause + a mergeable PR. 100 free investigations to start + 50 free/month, forever. The Problem Everyone is trying to build their own bug-fixing agent internally. The idea is obvious: point an agent at your alerts, let it fix the bugs. So people vibe-code a version over a weekend. But this is what actually happens: - the MCPs never authorize cleanly outside a local Claude Code setup, so the agent can't reliably reach your own services (DB, PostHog, etc.) without weird workarounds - it doesn't auto-trigger the moment a bug appears, so you're still the one noticing and kicking it off - it escalates everything (and token consumption explodes), or nothing, because there's no logic for when to pull in a human vs. handle it silently - it goes quiet while it works, so nobody in Slack knows what it's doing or whether to trust it - it has nowhere to actually reproduce the bug, fix it, and QA the fix, so the PR is a guess A few days later, the prototype gets abandoned, and you're back to copy-pasting stack traces into Claude. To solve this, we took the exact framework behind Superlog's bug-fixing agent (which normally runs on Superlog's telemetry), and pointed it at the stack you already run. That's Responder. lnkd.in/ehRWgHRD The Solution No new telemetry. You plug Responder into what you already have: - connect it to your Sentry/Datadog Slack channel - get your bugs fixed On every alert, Responder investigates with full context, filters out the noise, and for real issues replies in the thread with the root cause, the evidence, and a mergeable PR (if it's worth it). Prompts, memory, repo, and accesses are fully customizable, so you're building your debugging agent, not renting ours.

move fast. break things. superlog fixes them. we just 10x'd the free tier on superlog.sh one click to send telemetry from Vercel, Railway, Render, AWS, GCP + more. we instantly start investigating every error and hand you a PR with the fix. ship recklessly. we've got the cleanup.