Review a diff through a finder and a skeptical verifier
A two-pass code-review playbook: one model proposes possible issues, a separate verifier challenges each one, and you inspect the supported findings before making a change. Includes copy-paste prompts, a security taxonomy, exclusions, and a resolution-rate worksheet.
How it works
The exact pipeline, end to end.
How it actually works
Exactly what it does — and when.
A developer runs the Finder prompt on a risky diff, then sends each candidate through the Verifier. Several weak findings are rejected or narrowed. The developer checks the supported items against the actual code and tests before deciding which changes are warranted; the prompts do not guarantee that every bug is found.
Why it wins
Use it to…
Real ways people put The $0 AI Code-Review Playbook to work.
Review a risky diff
Run the finder prompt, then challenge every candidate with the verifier before acting on it.
Add a security-focused pass
Apply the taxonomy and exclusion list to code that handles input, access, secrets, or external calls.
Track useful findings
Record which findings led to accepted fixes instead of optimizing for raw alert volume.
What's inside
The shift it creates
- Staring at a blank prompt box
- Generic, inconsistent output
- Rewriting the same prompts daily
- Finder → adversarial verifier → confirmed-only
- Copy-paste Finder + Verifier prompts
- Security lens: full vuln taxonomy + exclusion list
Who it's for
Developers reviewing a risky diff
want a structured second challenge before accepting a model finding
Small teams without another review service
can use models they already configure
Security-conscious maintainers
need a taxonomy and exclusions to focus the pass
Questions, answered
Ready to put The $0 AI Code-Review Playbook to work?
Free download · no account needed · yours forever