diff --git a/content/blog/2025-02-26-bash-count-firefox-tabs.md b/content/blog/2025-02-26-bash-count-firefox-tabs.md
new file mode 100644
index 0000000..2bd8156
--- /dev/null
+++ b/content/blog/2025-02-26-bash-count-firefox-tabs.md
@@ -0,0 +1,43 @@
++++
+title = "Count your opened Firefox tabs with bash"
+date = "2025-02-26"
+updated = "2025-02-26"
+[extra]
+authors = ["Ydreniv"]
++++
+
+At any point in time, I have many Firefox tabs opened.
+I might even have reached a thousand.
+This feels like an issue I should address at some point, and I have already
+tried.
+But today, my goal is easier : count the number of currently opened tabs.
+In truth, it’s fairly easy : many extensions provide this information, in more
+or less granular details (number of windows, number of tabs in a specific
+window…).
+However neat this is, I want to get my total number of opened tabs printed in
+a Unix shell.
+Indeed, I ultimately would like to automatically graph this number over time.
+
+Before I started, I figured I’d need to find a session file, list the tabs,
+and get the count.
+Sounds fairly easy right ?
+Well it was not too hard in the end, but there were a couple bumps.
+The first is the format used to store the sessions.
+It’s compressed using LZ4, but with headers set by Mozilla, before LZ4 was
+fully standardized.
+[This StackOverflow](https://superuser.com/a/1363751) answer gives a bunch of
+tools to decompress the file.
+I’ve chosen to use [andikleen’s tool](https://github.com/andikleen/lz4json.git)
+as it worked out of the box easily.
+Once in possession of the extracted JSON, I’ve used `jq` to get the tabs’ IDs, and
+`wc` to count them.
+
+So with `./lz4jsoncat /home/<user>/snap/firefox/common/.mozilla/firefox/<profile-id>.default-release/sessionstore-backups/recovery.jsonlz4 | jq '.windows[].tabs[].index' | wc -l`,
+I’ve gotten my answer in a shell : 138 (yikes !).
+I’m using only one Firefox profile, but I suppose you could adapt this shell
+pipe to support more.
+Anyway, I’ll need to work on a cronjob to do it regularly, and an agent to
+collect these data (probably Prometheus).
+This would be one of the first metrics I’d collect for my "local computer
+monitoring center".
+