WEBAI KNOWLEDGE ATLAS · EDITORIAL PREVIEW

WebAI Studio & Lab✳

Interactive tools that run entirely in your browser, without an API or account.

WEBAI STUDIO & LAB · 15 MODULES

A clearer workspace with inspectable results

Choose a category, find a tool and open only the modules you need. Try the examples and inspect results. Calculations run locally, with limitations clearly stated.

Local No account Examples & results

Find a module

Direct access to all 15 tools

01Prompt builder 02JSON formatting 03Length and tokens 04WCAG contrast 05Agent approval 06Robots.txt 07JSON-LD data 08RAG chunking 09XML sitemap 10HTML checks 11URL inspection 12MCP schema 13JSON diff 14Evidence worksheet 15Markdown preview

All tools work locally

No text is sent to a remote service. No fake AI output is presented as model-generated.

01 / DESIGN

Prompt builder

Build a clear request and copy it to your tool of choice.

Expected role Goal Context and constraints Output format Editable result

02 / DATA

JSON validator & formatter

Check syntax, pretty-print or minify a JSON document.

Your JSON

No JSON checked yet.

03 / ANALYSIS

Text length estimator

Counts characters and words; token counts are a rough estimate only.

Text to analyze

0Characters

0Words

≈ 0Est. tokens

Simplified approximation (~4 characters/token), not a substitute for the model tokenizer.

04 / ACCESSIBILITY

Contrast checker

Compare two colors using the WCAG contrast formula.

TextBackground

WebAI · Aa 123

—Ratio

—WCAG AA

Contrast calculation is not a full accessibility audit.

05 / SAFETY

Agent approval simulation

Sensitive actions stay blocked until explicit approval. This demo contacts no service and makes no booking.

Task: prepare a fictional bookingAwaiting approval

No action performed.

NEW TOOLS · LOCAL ANALYSIS

Turn concepts into practical checks

Ten additional modules execute in your browser only. No text uploads, no remote AI and no claim of complete validation. Each checker explains its own limits.

06 / ROBOTS

AI crawler policy

Compare search and training crawler preferences. robots.txt does not enforce access control.

Path to disallow Crawler

The file will appear here.

OpenAI crawler documentation ↗

07 / JSON-LD

Structured data builder

Describe a resource without inventing invisible properties. This builder does not test search engine eligibility.

Type Exact title Public URL (HTTPS) Author or organization

—

08 / RAG

RAG chunking explorer

Visualize deterministic word chunks with overlap. No embeddings or vector index.

Corpus text

Words per chunkOverlap

—

09 / XML

XML sitemap inspector

Parses XML and some Sitemap protocol elements, without fetching URLs or checking indexing.

sitemap.xml content

—

10 / A11Y

Basic HTML checks

Checks h1, main, missing image alt attributes and unlabeled inputs. Not a replacement for WCAG or human testing.

HTML source

—

11 / URL

URL inspector

Find host, path, query and fragment. No network request or trust verification.

URL

—

12 / MCP

MCP tool descriptor builder

Draft tool metadata and a JSON input schema without running or exposing a server.

Technical name Action description Required parameter

—

13 / JSON

JSON document comparison

Compare paths and values in two JSON documents without changing either.

JSON A JSON B

—

14 / SOURCES

Citation review worksheet

Record a claim, source excerpt and date; you decide whether evidence supports it. No automatic semantic verification.

Claim Original excerpt Source URL Date checked Human judgment

15 / MARKDOWN

Safe Markdown preview

Render headings, paragraphs, lists and quotes from plain text. This is not a complete CommonMark engine.

Markdown

—