Blog
Notes from the gate
Running AI agents against real repos, on self-hosted hardware, behind a push gateway. Field notes and honest numbers from building nimblegate.
-
Parallel AI coding agents: get alerted when two edit the same files
Parallel AI coding agents collide at merge time, even in separate git worktrees. See two branches touching the same files while the agents still work.
-
Running AI coding agents in parallel on one repo
Parallel coding agents on one repo are fast until two edit the same file. What I measured, fixes that work today, and how to spot overlaps early.
-
I loop AI agents on my repos. Here's an easy way to see what they push.
Running coding agents in a loop is fast and fun. The hard part is seeing what they push and stopping a bad push before it lands. Why hooks were not enough.
-
Running a local AI model smoothly on a $214 used office machine
A 35 W i5-10400T with no GPU runs a 35B-parameter model at ~8 tokens/s, with web search, RAG, and a queryable git gateway over MCP. The numbers, honestly.