Pinned Tweet
The open-source code is MIT licensed, free forever: github.com/saxonnicholls/sup…
Remember Log4Shell from 2021? Log4j 2.0-beta9 through 2.14.1 allowed remote code execution. Most teams didn't even know they had it; it was buried in transitive dependencies. The scramble wasn't patching, it was finding out what versions were actually running. We solve this for $7/month
1
44
Remember Heartbleed from 2014? OpenSSL 1.0.1 through 1.0.1f leaked server memory to anyone who asked. The fix was simple: upgrade. The hard part for most teams was answering "which of our servers are running which OpenSSL?" We solve this for $7/month. Largest version mismatch database in the world
Remember GitLab from 2017? GitLab accidentally deleted a production database, then went to restore from backups. There weren't any usable ones. The backup job ran pg_dump 9.2 against a Postgres 9.6 database and failed quietly, and the failure alerts never arrived. Mismatched tools, silent errors, lost data. We are the largest version mismatch database in the world. Yours for $7/month
Remember Equifax in 2017? The Apache Struts patch was out in March. Equifax was still running the old version in May. 147 million people's data was exposed. "What version is actually running in prod?" is a question worth answering every day. @superloghq has the largest version mismatch database in the world for $7/ month ….
Remember Ariane 5 in 1996 ?
Ariane 5 reused Ariane 4's guidance software. Faster rocket, bigger numbers, a 64-bit float crammed into a 16-bit int. Overflow. Self-destruct 37 seconds after launch. The code worked, just not for this version. Makes $7/month look like good value
Remember Knight Capital from 2012?
One of 8 servers didn't get the new deploy. A reused flag woke up dead code on that one box. 45 minutes later, Knight Capital had lost ~$460M. One version mismatch. One server. One company gone. Makes $7 per month look pretty cheap….
2
26
We are building the world's largest database of version mismatches - software, drivers & hardware to save you time and anguish. Versions often drift and silently break things - with super-log no more. Version interactions can be even more insidious.
1
20
5/ It's not only for whoever reads it afterward. `search_logs`, `tail_logs`, `wait_for` — the same MCP tools a person uses in the console are what an agent calls too. One agent can check what another just did
before it acts, instead of guessing or colliding.
1
6
6/ We build super-log this way ourselves — several coding agents sharing one repository, one bench, one ordered log of everything they did.
One install, free, MIT:
`claude mcp add super-log -- npx -y @super-log/mcp`
super-log.com
1
10
3/ Two of them touch the same deploy within moments of each other. Reconstructing what happened means stitching together clocks that don't agree, from logs that were never meant to be read together.
1
1
2/ Five agents. Five different things being built: a scraper, a mobile build, a k8s rollout, a warehouse job, a billing service. Each one logging into whatever it was handed — its own stdout, a file nobody tails, or
nothing at all.
1
1
10
1/ A swarm doesn't fail because one agent breaks. It fails because nobody can tell, afterward, what all of them actually did — in what order, on what machine, to what end.
1
1
12
@superlogHQ is THE key enabling technology for true agentic debugging - get a lot more efficiency from @claudeai @OpenAI @grok @GeminiApp - present consolidated logs with the code and the errors - @code plugin now available
1
5,241
131,930 MCP servers on the registry. super-log is in the top 10% since launch week. AI debugging has just arrived.
Pasting a stack trace into a chat window doesn't count — that's you doing the debugging and the model doing the typing.
One hub, every stream on the machine, in the order it happened.
github.com/saxonnicholls/sup…
1
34


