Worked examples / 3 October 2026

Check launch data before you publish

Clean a small export, keep a campaign link consistent, and check whether your text is readable. Each tool produces an inspectable result from your inputs. No result is a traffic forecast.

Make a small export readable

Suppose an automation returns campaign counts as JSON. Paste a flat array into the converter, inspect its CSV preview, and open the downloaded file in your spreadsheet. Columns are the union of the supplied fields. Missing properties and null values become blank cells. Nested data is rejected rather than flattened ambiguously.

Up to 100 records, 30 columns and 12 KB. Keep large identifiers as strings. Formula-like cells receive a leading apostrophe, so negative numbers also need review before numerical analysis.

Example result

channel,visits
newsletter,24
directory,12

Run the same check through MCP

Call toolfound_json_to_csv with:

{
  "json": "[{\"channel\":\"newsletter\",\"visits\":24},{\"channel\":\"directory\",\"visits\":12}]"
}

The result contains ok, human-readable sections, structured data and the canonical result URL. MCP inputs are sent through your AI client to Toolfound; browser forms compute locally.

Try this check

Remove repeated rows without merging different records

A combined listing export may contain the same row twice. This tool retains the first complete row and removes later exact matches. It does not merge people, infer identity or choose a preferred campaign. Case, spaces and every column matter.

Headers plus up to 100 rows, 30 columns and 12 KB. Quoted commas, quotes and line breaks are supported. Selective-column and fuzzy matching need a different workflow.

Example result

3 input rows → 2 output rows; 1 duplicate removed.

Run the same check through MCP

Call toolfound_csv_deduplicator with:

{
  "csv": "channel,visits\nnewsletter,24\nnewsletter,24\nnewsletter,25"
}

The result contains ok, human-readable sections, structured data and the canonical result URL. MCP inputs are sent through your AI client to Toolfound; browser forms compute locally.

Try this check

Fix campaign naming before sharing

Choose normalize when you want a consistent campaign label: Launch News becomes launch_news. Choose remove when you need a plain reference link. Neither operation rewrites historical analytics. The destination is not fetched, so the result does not establish that the link works.

Only the five conventional UTM values are normalized. The remove action deletes all utm_ parameters. Unrelated parameters and the anchor remain; recognized credential fields and duplicate UTM keys are rejected.

Example result

https://example.com/?utm_source=launch_news&ref=partner#pricing

Run the same check through MCP

Call toolfound_campaign_url_cleaner with:

{
  "url": "https://example.com/?utm_source=Launch%20News&ref=partner#pricing",
  "mode": "normalize"
}

The result contains ok, human-readable sections, structured data and the canonical result URL. MCP inputs are sent through your AI client to Toolfound; browser forms compute locally.

Try this check

Check the pair your reader will actually see

Use the text and background hex colors from your design. #777777 on white produces about 4.48:1, below the 4.5:1 AA threshold for normal text. The decision uses the unrounded ratio. Large text has a lower 3:1 AA threshold; it means at least 18pt, or 14pt bold.

Opaque solid colors only. Images, transparency, interaction states, focus indicators and keyboard navigation require separate inspection. A color-pair result is not a full accessibility assessment.

Example result

AA normal: fail. AA large: pass. AAA normal and large: fail.

Run the same check through MCP

Call toolfound_color_contrast with:

{
  "foreground": "#777777",
  "background": "#ffffff"
}

The result contains ok, human-readable sections, structured data and the canonical result URL. MCP inputs are sent through your AI client to Toolfound; browser forms compute locally.

Try this check

Keep the original evidence

Keep a copy of the original export. Inspect the output before replacing any source file. These tools do not store an account history, verify a vendor, or automatically publish anything.

Contrast thresholds and the large-text definition follow the W3C contrast guidance.

Building a tool of your own? Submit it to Toolfound for free. Verified makers can review Plus campaign features separately.