Frequently asked questions

Three options. Paste a public GitHub repo URL and we read it on our server, then delete it. Run a private repo or a local folder in browser memory, so raw source handling follows the selected data path; configured external tools may receive scoped derived evidence. Or use the CLI to point it at code locally or in CI. Whichever fits your comfort with privacy.

It depends on the selected path. Public repositories use isolated temporary server jobs with terminal cleanup; private GitHub uses the documented transient relay; local folders and ZIP files are read in browser memory; CLI analysis runs locally. A configured remote model may receive derived evidence.

Three artifacts, with available evidence references: (1) a founder runbook to operate and extend your app; (2) a ranked code-verified fix-list to hand to your AI; and (3) an end-user guide for the people who use your app. The runbook and fix-list are free; the end-user guide is proposed and not currently for sale.

The end-user guide is written for the people who use your app — plain-language help, not code. That's different from developer docs, which are for engineers. Notably, we generate the end-user guide from your code — see end-user vs developer docs.

The founder runbook and code-verified fix-list are free. The end-user guide plus sharing (share link and access code, in-app Help widget, portable export) is proposed and not currently for sale. Full details on the pricing page.

The proposed commercial plan would add the end-user guide plus a share link with an access code, an in-app Help widget you can embed for your users, and a portable export you can take anywhere. It's the difference between having a guide and getting it in front of your users cleanly.

Yes, and there's no catch. The founder runbook and code-verified fix-list are genuinely free — the two artifacts that help you operate your app and fix real problems. The proposed commercial tier would add the end-user guide and the sharing tools around it. You can get real value without paying a cent.

It's a continuous audit you wire into your CI pipeline. On every push it re-reads your code, grades the build, and flags what got worse since the last scan — a regression signal. It runs in your own infra, so raw source handling follows the selected data path; configured external tools may receive scoped derived evidence.

Codacy and SonarQube grade code quality and security (static analysis, secrets, dependencies, quality gates). The Explain My Build audit grades documentability and docs-drift — whether your build can still be explained and what changed for the worse. They're complementary; run both if you're CI-comfortable.

Supported output claims retain evidence references and unsupported conclusions are marked unknown. When Explain My Build can't verify something against your code, it flags it or leaves it out rather than inventing a plausible answer. A shorter honest guide beats a longer confident wrong one. See limitations.

No. It reads code and can flag issues visible in code — like a loose row-level security rule or a committed secret — but it doesn't attack your running app or replace a real security review. For regulated data, treat it as one input and get a qualified human review. See limitations.

Projects from many code-producing tools can be analysed when supported source is available — Lovable, Bolt, v0, Replit, Cursor, Windsurf, Base44, Claude Code, Gemini, and more. We read the code, not the tool.

No. The whole point is to turn your code into plain English for a non-technical founder. You paste a URL or point it at your folder, and you get readable guides and a clear fix-list. You don't read code; you read the explanation of it. See get a help guide without reading code.

When real money, health, or legal data is at stake, when you've hit the same wall three times, or when uptime becomes business-critical. The runbook and fix-list don't replace a developer for high-stakes work — but they make the handoff far cheaper. See who maintains my AI app.

No. Guide Pass and Pro are proposals, and payment and upgrade controls remain disabled pending independent legal review. See pricing for current status.