Home › About
About JSONSalvage
Direct answer: JSONSalvage is operated by Abdul Hannan, AH Digital Studio, Pakistan. It exists for one job: recovering broken JSON — with every change explained, and with your data never leaving your browser.
Why this site exists
Broken JSON is a daily tax on developers and automation builders: LLM responses cut off mid-stream or wrapped in prose, exports with a few destroyed lines among thousands of good ones, configuration files in a dialect that is almost JSON. The existing choices were a formatter that tidies silently, or a repair tool that uploads your payload — payloads that routinely contain keys, tokens, and customer data — to someone else's server.
JSONSalvage takes the opposite position on both counts. Processing happens entirely in your browser, on your own device; there is no server copy of anything you paste. And repair is deterministic with a published account of itself: a Repair Report that lists every change, so a fix can be reviewed like a diff instead of trusted like a rumour.
What it will not do
- It will not invent data. A missing value or a truncated tail is reported, never completed with a guess.
- It will not silently change data. Duplicate keys are preserved and flagged; numbers keep their exact source text.
- It will not claim to be something it is not: no "AI repair" (the repair follows a fixed, deterministic set of rules), no certification of MCP files (the registry decides), no invented usage numbers anywhere on the site.
The tools
JSON Repair is the flagship. NDJSON Salvage recovers line-delimited files record by record. Truncated JSON Recovery returns the largest honest prefix of a cut-off document. The MCP server.json Validator is a supporting tool for registry publishers, checking against the official schema version printed on its page. The methodology guide explains the repair rules in full.