The GenomeComputer Company
Back to tutorials

Tutorial

Use Codex with your .genome

Use a local coding agent to inspect your .genome files, search annotations, write reports, and build repeatable exploration workflows.

7 minCodex, Claude Code

What you will learn:

You will know how to use Codex or Claude Code to work with your .genome locally while preserving file references and evidence.

Codex and Claude Code are useful when you want your genome exploration to happen inside a local project folder. Instead of using a general chat interface, you can put your .genome bundle in a workspace and ask an agent to inspect the actual files, search across annotations, write scripts, and create reports you can keep.

This is especially useful for repeatable workflows. A chat answer is easy to lose. A local workflow can create markdown reports, search scripts, reusable prompts, and small tools that you can run again when your .genome is updated or when Genome Intelligence adds new research.

Local workspace setup

  1. Create a folder for genome exploration.
  2. Place your .genome bundle or extracted .genome files inside it.
  3. Open the folder in Codex or Claude Code.
  4. Ask the agent to list files and identify file types.
  5. Ask it to search only after it has mapped the bundle.
  6. Ask it to write findings to markdown so you can keep the output.
  7. Reuse the same workspace for future topics.

Workspace orientation prompt

Inspect this .genome bundle in the current workspace. First list the files and identify which are best for AI-readable summaries, variant lookup, source review, prompt reuse, and technical review. Do not analyze a trait yet. Create a file map and wait for my next question.

Search and report prompt

Search this .genome bundle for [gene, variant, pathway, trait, or paper]. Return a markdown report with file references, exact matches where available, relevant annotations, source fields, and follow-up questions. Separate direct file evidence from broader interpretation.

What Codex can help with

  • Search across many files without manually opening each one.
  • Create a table of genes, variants, sources, and notes for a topic.
  • Write scripts that extract repeated fields from your .genome.
  • Compare two exploration reports over time.
  • Turn a Genome Intelligence prompt into a repeatable local workflow.

Make the output reusable

  1. Ask the agent to save the report as markdown.
  2. Ask it to include the prompt that produced the report.
  3. Ask it to include file paths and source fields.
  4. Ask it to list what it did not inspect.
  5. Ask it to propose a next run with one improvement.

A good local workflow feels less like a one-off chat and more like a research notebook. Your .genome is the data layer, the AI agent is the assistant, and the saved reports become your memory of what you explored and how.