Contact PromptOptBase

Send email to the address shown above. We do not operate a ticket portal, live chat, or social inbox for support — email keeps a clear thread and lets you attach screenshots when helpful.

No sign-up is required to use PromptOptBase, and contacting us does not create an account. We never receive the prompt text you type into tools unless you paste it into your message. Tool processing runs in your browser; our servers do not store your drafts during normal use.

A useful message saves back-and-forth. Please include as much of the following as applies:

• The full page URL where you saw the issue (or which tool you were using)
• Browser name and version, operating system, and the site language you had selected
• For tool bugs: the inputs you entered (or a redacted sample), what you expected, and what actually happened
• Screenshots or a short screen recording if the problem is visual or hard to describe in words
• Whether the issue reproduces in a private/incognito window with extensions disabled

You do not need to send your entire prompt if only one section misbehaved — a minimal example that reproduces the bug is enough. Redact customer names, credentials, or confidential business data before sending.

We read messages on a best-effort basis. PromptOptBase is a small, independent project — not a staffed 24/7 support desk.

• No guaranteed response time and no service-level agreement (SLA)
• Urgent production incidents on our hosting may get faster attention; general how-to questions may wait longer
• We may ask for a follow-up detail or a clearer reproduction step before we can act
• Silence usually means we are backlog-heavy, not that your report was ignored — a polite bump after a couple of weeks is fine

We cannot troubleshoot third-party chat products (ChatGPT, Claude, Gemini, etc.) or predict how a model will respond to your pasted prompt.

These are the reports and questions that help us improve the site:

• Rewrite or assembly output that contradicts our published methodology
• UI bugs: broken layout, copy button failures, missing sections, or incorrect routing between tools
• Privacy questions about hosting, analytics, cookies, or what we do and do not collect
• Template corrections — misleading placeholders, wrong channel constraints, outdated labels
• Translation or locale issues: awkward phrasing, mixed languages, or broken text in a non-English UI
• Feature suggestions that fit our local-first, rule-based scope (we may decline, but we read them)
• Accessibility friction you noticed on a static page or tool form
• Broken or outdated links in guides or template descriptions

To set expectations, we generally cannot assist with:

• Personalized prompt coaching, ghostwriting, or reviewing your full draft for quality
• Jailbreak, bypass, or "make the model do X no matter what" requests
• Guaranteed model outputs, ad approval, SEO rankings, or business outcomes from pasted prompts
• Recovering prompt text you entered only in your browser — we have no remote copy unless you send it
• Legal, compliance, medical, or security certification of your customer-facing copy
• Integration support for vendor APIs, enterprise SSO, or custom deployments we do not ship

We improve structure and clarity with deterministic rules. You still own facts, policy fit, and the decision to send a prompt to any model.

If you believe PromptOptBase infrastructure is being abused — for example, phishing pages on our domain, malware links injected into static content, or sustained automated scraping that threatens availability — email us with evidence.

• Include timestamps, URLs, and any logs or headers you can share safely
• Describe the impact: data exposure, impersonation, downtime, or deceptive content
• Do not include exploit payloads or live attack instructions in the email body

We do not accept unsolicited penetration-testing requests or "please try to hack us" challenges. Coordinate any authorized security review in writing before testing.

Many questions are already answered on the site. Checking these first often resolves the issue faster than waiting on email:

1. Read the methodology page — it explains what our rewrite and assembly rules do and do not claim
2. Open the relevant guide for the technique or channel you are using
3. Scan the tool page itself for inline notes, constraints, and FAQ-style hints near the form
4. Try a private/incognito window to rule out extensions or stale cached scripts
5. Switch locale or browser once to see if the problem is translation-specific or environmental

If something on a tool page disagrees with methodology, treat that as a bug and tell us both URLs in one message.

For data handling, acceptable use, and how we maintain rewrite rules, see the pages below. Citing the section that confuses you helps us reply precisely.

Privacy · Terms · Methodology