glunty 8 tools in this category. Each runs in your browser.
AI-augmented tools forward your input to Anthropic Claude for processing and return a result. They handle the cases where deterministic local processing cannot: explaining what a regex actually matches in plain English, walking through a SQL query clause by clause, breaking down a bash command before you run it, decoding an Excel formula nested calls, sharpening a vague prompt into a more specific one, rewriting an email in a chosen tone without inventing facts, sharpening a resume bullet without fabricating numbers.
The privacy model: glunty does not log or store the input. Anthropic processes the input for the duration of the request under their data-usage policy (no training on API inputs by default). Where the tool deals with potentially-sensitive content (resume bullets, email drafts, SQL queries with literal values), the per-tool FAQ flags the consideration so you can strip secrets before pasting.
Free tier limits are per-IP per-day and vary by tool: 10 explanations per day for short-input tools, 5 for longer-input or vision-model tools. The caps keep the service genuinely free without enabling automated abuse that would force a paywall. For higher volume, the Claude API directly is straightforward and inexpensive.
Output reliability: AI explanations are usually right but not guaranteed. For load-bearing use (security filters, financial models, code in production), verify the output against authoritative sources or a human expert before relying on it.
The ideas behind these tools, in plain English.
A large language model predicts the next token from the preceding text, one piece at a time, based on patterns learned from training data.
Seeing it as sophisticated prediction rather than reasoning explains both its fluency and its failure modes.
A model generates plausible-sounding text even when it has no grounding, producing fabricated facts known as hallucinations.
Confidence is not evidence, so anything factual from an AI needs verification before you rely on it.
Specific prompts that state the goal, format, constraints, and examples steer the model toward a useful answer.
Vague requests get vague output, so a few extra words of context often replace several rounds of correction.
Text you send to a hosted model leaves your machine and may be logged or retained by the provider.
Pasting secrets, keys, or personal data risks leaking them, so strip or mask anything you would not put in a support ticket.
Plain-English breakdown of any regex.
Walkthrough for any language.
Clause-by-clause walkthrough.
Token-by-token shell breakdown.
Function-by-function explanation.
Sharpen a vague prompt.
Rewrite in formal, friendly, or firm tone.
Rephrase only; never fabricates.