CARBON — Engineering ConfidenceUser Guide
CHAPTER 37 / 60 Pro

/carbon-review

Independently challenge AI-generated code, an AI behavior change, an eval suite, or release evidence for hidden assumptions, unsafe interactions, weak measurements, and false confidence.

Jason · AI-generated codeProof & release confidence

When to use it

You want a fresh challenge to AI-generated work or supporting evidence.

Before you start

Provide the diff, intent and current checks. Trace plausible shortcuts, missing boundary handling and weak assertions. Make suspected defects actionable without presenting static inference as a reproduced failure.

Try this in your coding agent.

/carbon-review Review the AI-generated changes in src/auth and identify missing tests.

Replace sample paths with your project. Slash names are shorthand; your host may use a plugin prefix or a skill picker. Invocation help →

ILLUSTRATIVE WALKTHROUGH

AI-generated authentication diff

An expanded teaching example, not a recorded or live execution. The report stays visible while the replay explains how to read it.

  1. 01Read context
  2. 02Investigate
  3. 03Show evidence
  4. 04Report

Coding-agent conversation

Read context

I’m reviewing ai-generated authentication diff. I’ll keep the scope and permissions explicit.

Investigate

Follow the weakest evidence link and record what remains blocked.

Show evidence

Here is the evidence trail to inspect: Diff trace: missing expiry value → default truthy branch → proposed boundary test.

Report

A plausible fallback bypasses the intended expired-session check. Static suspicion needs reproduction before calling it a verified bug.

CARBON

Example report

Illustration
Inspect the identity boundary
Inspect the identity boundary — TestBucks teaching screen

TestBucks course screen with illustrative analysis. The screenshot supplies context; the demo does not establish a newly observed bug.

  1. User action
  2. Identity / validation
  3. Allowed outcome

Illustrated outcome

A plausible fallback bypasses the intended expired-session check.

Evidence to inspect

Diff trace: missing expiry value → default truthy branch → proposed boundary test.

Still untested / not established

Static suspicion needs reproduction before calling it a verified bug.

Ready. Read the full example or play its steps.

What a useful result contains

A bounded claim

For ai-generated authentication diff, state the target, revision, environment and actual work performed. Keep planned, executed, blocked and untested work separate.

An inspectable trail

Diff trace: missing expiry value → default truthy branch → proposed boundary test. In a real run, retain the source, trace, screenshot or fixture behind the conclusion.

The qualification

Static suspicion needs reproduction before calling it a verified bug.

The next decision

Decide whether the evidence calls for a focused check, a reviewed fix or more context. Do not convert a plausible example into a verified finding.

Adapt the example to your project

  1. Replace ai-generated authentication diff with a real feature, path or decision in your project.
  2. State the expected rule and its source. Include safe fixtures, constraints and a budget where appropriate.
  3. Ask for exact evidence and remaining gaps. Approve consequential actions separately.
/jay Review the /carbon-review result. Separate observed evidence, assumptions and untested scope, then recommend the next useful check.

Where to go next

/carbon-test →/carbon-skeptical-review →

Content checked against the local command definition: skills/carbon-review/SKILL.md. Examples adapted from the homepage’s TestBucks walkthroughs. Availability is edition- and host-dependent.