BASHMATICA!

Hard-Won Insights Into DevOps Automations & Infra AI

Articles

llm-agents

Sep 1, 2026

•

12 min read

GitHub's August Was Fire

Literally on fire - outages, massive instability, and a fun new way to steal secrets

code-review

Aug 25, 2026

•

12 min read

Uncle Bob Stopped Reading the Code

He barely looks at the code anymore and reports 5x. What matters is whether the things he looks at instead can fail the run, or only nod at it.

Bobby R. Goldsmith

llm-agents

Aug 18, 2026

•

11 min read

Every Autonomous Run Needs a Fuse

An unbounded agent that hits a loop doesn't stop when the work is done; it stops when the bill arrives. Put a fuse on it that trips first.

Bobby R. Goldsmith

evals

Aug 11, 2026

•

11 min read

Evals Are Regression Tests for a Model That Moves

Your code is frozen until you change it. The model behind it isn't, so the only thing standing between a silent regression and your customers is a graded set of cases the change has to pass first.

Bobby R. Goldsmith

model-drift

Aug 4, 2026

•

10 min read

Temperature Zero Is Not a Lock

You ran the same prompt twice at temperature zero and got two answers. The knob you trusted to freeze the model was never a lock, so stop pinning the model and start pinning the pipeline.

Bobby R. Goldsmith

code-review

Let Your AI Reviewer File a Brief, Not a Verdict

Jul 28, 2026

•

10 min read

Let Your AI Reviewer File a Brief, Not a Verdict

An AI reviewer reads a diff well, but the moment its green check counts as the approval, you've handed the merge key to something an attacker can steer with a sentence.

Bobby R. Goldsmith

BASHMATICA!

Articles

GitHub's August Was Fire

Uncle Bob Stopped Reading the Code

Every Autonomous Run Needs a Fuse

Evals Are Regression Tests for a Model That Moves

Temperature Zero Is Not a Lock

Let Your AI Reviewer File a Brief, Not a Verdict

Fortify with Bashmatica!