🧪 AI Build Sanity Check

Catch visible signs that an AI-built app is not finished

Run a local, read-only Python scanner before a demo, review, or launch. Each finding reports a rule ID, severity, file, line, and plain-English next step. The scanner does not modify or execute the app.

Six defined checks

• Build-marker terms such as TODO, FIXME, “not implemented,” and NotImplementedError.
• Return lines containing configured fake, mock, sample, dummy, or placeholder terms.
• Python except blocks followed by pass and same-line empty JavaScript catch blocks.
• Configured placeholder strings in supported source, config, and .env files.
• Return lines with success, ok, or status near a supported route marker.
• Route-adjacent empty or throwing handler markers as advisory review prompts.

What you receive

The package includes the standard-library scanner, editable JSON rule table, remediation snippets, audit checklist, explanations, clean and risky fixtures, and a test file. The fixture test passes with 11 findings: 10 deterministic and 1 advisory.

Know the boundary

This is line- and term-based static analysis, not proof of unfinished behavior. It can flag intentional TODOs, demo code, or legitimate success responses and can miss patterns written differently. It does not run the app, test behavior, or assess security.

Handle .env deliberately

The scanner reads supported local files, including .env files within the chosen path, to find configured placeholder terms. It reports file, line, rule, severity, and a generic message—not the matched line or value—but you should scope the run away from .env if you do not want that file read.