New writing appears in the feed as soon as it is published.
-
Firecracker on WSL2, without root
How I run each coding agent in its own Firecracker microVM nested in WSL2, with no root at run time: the stack, what it costs, and ten stumbling blocks.
-
Credential proxies fail open in the details
Where credential proxies for AI coding agents fail open: Docker's and microsandbox's advisories of autumn 2026, three decisions a proxy makes, and what d0 does at each.
-
EmbeddingGemma in Go, tokenizer and all
How “ask teblo” matches questions with EmbeddingGemma in plain Go: the models measured, the tokenizer, the one output that matters.
Follow along: https://mko.run/feed.xml