Skip to content

Bruce Hart

Latest

Feb 15, 2026 2 min read

I Put Codex Spark in My .bashrc and It Changed My Command Line

I added a tiny Bash helper that uses Codex Spark to turn plain-English intent into shell commands. The speed makes it practical for everyday one-offs, with a few safety guardrails.

I stopped context-switching to docs for weird shell one-offs.

I added a small script to my .bashrc that lets me describe what I want in plain English and get back a Bash command from Codex Spark. Spark is really fast and the Codex models in general are wizards of the command line.

The win is not that I suddenly forgot grep or awk. The win is loop time. Spark is fast enough that I can stay in terminal flow, ask for the command, sanity-check it, and move on.

Fast feedback changes what gets automated

Most command-line friction is tiny but constant.

You know the task, you kind of know the tool, and then you lose five minutes remembering one flag, quoting rules, or regex syntax. That usually kills momentum more than it kills correctness.

With Spark in the loop, the threshold drops. I now automate small tasks I used to do manually because the assistant response is quick enough to feel native to shell work.

Prompt quality now matters more than syntax memory

I noticed a shift in what skill is rewarded.

I spend less time memorizing exact sed spellings and more time writing precise intent: input shape, output format, constraints, and edge cases. When the prompt is specific, command quality improves a lot.

That is a good trade for me. I would rather be great at describing outcomes than great at remembering obscure flag combinations.

Safety is a workflow, not a warning label

Command generation is useful, but shell is sharp.

My default pattern is simple: inspect first, run second. For destructive operations, I ask for a preview variant or dry run before the real command. I also tend to split giant one-liners into readable steps so I can reason about them.

Spark helps with this too. You can ask it to explain each segment of a pipeline before execution, which makes review much faster.

Tiny helpers compound over time

This .bashrc helper is small, but it has already changed my day-to-day terminal behavior. I use the command line for more tasks, with less hesitation, because the lookup tax is lower.

Source

The script referenced in this post: https://gist.github.com/brucehart/81ee6a5c45ce4ee784d4e4711c6f792d

Read the full piece

More articles

AI Feb 13, 2026

Codex Spark: Speed vs Depth

Codex Spark is fast enough to change what I bother automating. The trade is a little less thoroughness on harder tasks, so I now use a two-model workflow.

4 min read
AI Feb 4, 2026

Unlocking More with Codex Skills

Treat Codex skills like evolving capabilities, not static tools. When you refine them, script the busywork, and review logs for gaps, Codex stops being a helper and starts feeling like a compounding system.

3 min read
coding Jan 17, 2026

When Coding Gets Cheap

LLMs are approaching top programmers; the bottleneck shifts from syntax to intent, Jevons Paradox likely drives more software, and software IP moats move beyond code.

5 min read
AI Jan 13, 2026

Chrome, the Clipboard, and Two Buttons

A practical AI workflow: use Gemini to generate one-off console scripts, and use small TamperMonkey buttons to turn web pages into clean Google Sheets rows and predictable Google Drive receipt filenames (Walgreens + HSA receipts case study).

6 min read
coding Jan 9, 2026

Web Tools: my tiny-tool factory on Cloudflare Workers

I built a single Cloudflare Worker that serves a suite of small web tools. Codex helps me spin up new ones fast, and when a tool needs login + persistence, I bolt on Google OAuth and D1 without turning the whole project into a full app.

6 min read