2.197 gb over 8,518 calls on dev, 95% of codenaya's db bandwidth gone on just listing files 😭 github import was writing one file at a time and every write re-read the whole growing project, content included
merged the fix: batch text file writes + pause the file list during import, test harness went from 1,578.8 mb read to 1.43 mb for 2,000 files (not prod numbers)
https://t.co/aCATL6pg4g
Meet Oh-My-Hermes
I burned 200B tokens tuning my Hermes Agent setup, and now you can get all of it with a one-line install.
It comes with 150+ skills, a long-term memory system that lasts twice as long, graph engineering, and memory-management skills. It also includes subagent routing across model chains, plus optimizations like parallel tool calling and evals.
You don't need to do much, though. Just install it and give it a task. Everything is optimized to route automatically. Hermes Agent is excellent, and I love this technology.
I put a lot into making it really good, then open-sourced it so you can enjoy it right away. Meet oh-my-hermes.
GitHub: https://t.co/g1exHwWK5E
#hermes
in the last 24 hours this rack
- handled 6.77T tokens
- p50 TTFT was 1.16s
- p50 TPS was 143
unfortunately p95 isn't looking great because one of the trays is flakey
but as soon as we get it replaced...this is gonna be amazing
@Cloudflare d1 undelete now rejects databases deleted more than 7 days ago. the undo button has an expiry date, put it in the incident runbook
https://t.co/fGO9HJXL65
who's building devtools or ai tools rn, and what's the one technical problem holding up your next release? drop the project + blocker, way more interesting than another follow train
@github cli 2.103 lets you opt into expiring tokens with gh auth login --expiring. nice, but scripts that stash the token forever still need fixing
https://t.co/GBjEtWSvqo
@Cloudflare cut ai review costs 68%, but lowering effort also lost real performance findings. cheaper tokens aren't a win if the bugs get a free pass
https://t.co/DDLSLL2ePu
queue length without oldest job age is half a dashboard. 10 jobs waiting since yesterday is a different problem from 10 jobs waiting since a second ago
@tan_stack ai 1.0 makes the core stable but keeps the new agent harness experimental. upgrade client and server together, the version number isn't a blanket blessing for every package
https://t.co/xQvbxp9dyi