Review Recommendations

cortex review is a standalone recommendation surface for review work. It is not the same thing as the agent-loops workflow.

The job of cortex review is:

Suggest which review-relevant skills should be loaded before you perform a review.

What It Is Not

cortex review does not run the full agent-loops review/test-audit loop for you. agent-loops is a larger implementation discipline workflow. cortex review is a composable helper you can use before review work begins.

Core Command

cortex review --dry-run
cortex review --context debug
cortex review --context feature --context security

The command takes optional context signals and recommends review-relevant skills based on what kind of review you are about to do.

When To Use It

Use cortex review when:

you are about to review a diff manually
you want Cortex to suggest the right skills before reviewing
you are preparing a review-oriented handoff for another agent or model

It is especially useful when the review mode is not obvious from the diff alone and you want a nudge toward the right review discipline.

Practical Workflow

# Ask for review-oriented skill recommendations
cortex review --context security --context api

# Build skill context for the session if needed
cortex skills context --no-write

Then load the suggested skills or use their generated slash-command aliases.

Common follow-on skills include:

/ctx:requesting-code-review
/ctx:doc-claim-validator
/ctx:security-testing-patterns
/ctx:python-testing-patterns

Relationship To Agent Loops

Keep the distinction clean:

cortex review recommends review-relevant skills
agent-loops is the full implementation workflow with review, test audit, lint, and commit gates

That means cortex review can be used on its own, or it can be one piece of a larger composable workflow.

Agent Loops – full implementation discipline workflow
Skills – inspect or rate the skills being recommended
Multi-LLM Consult – take review context to another model if needed