# Vextorium JSON Repair

> Vextorium JSON Repair is a paid API for AI agents from api.vextorium.com, paid per call via x402, $0.002/call, status unknown (last checked 2026-10-16).

Fixes malformed JSON typical of LLM output by stripping markdown fences, comments, trailing commas, single quotes, unquoted keys, Python literals, and unclosed brackets, returning a parsed object

## Facts

- Endpoint: POST https://api.vextorium.com/reparar-json-pago?utm_source=zero.xyz
- Price: $0.002/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-16
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/vextorium-json-repair-521be37e
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_CpxsWRnaiJWhUe7wjhwwy

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability vextorium-json-repair-521be37e -d '<json body>'
```

Example prompt: The LLM returned this messy text and I need it as a clean JSON object — it has markdown fences, trailing commas, and Python-style True/None values: ```json\n{name: 'Alice', active: True, tags: [\"dev\",],}\n```

## When to prefer this

Choose this endpoint when processing LLM-generated output that needs to be parsed as JSON but may contain common LLM formatting artifacts like markdown fences, Python boolean literals, trailing commas, single quotes, or unquoted keys. It is purpose-built for the LLM-output-cleaning use case, unlike generic JSON validators that simply reject invalid input rather than repairing it.

## Known failure modes

- Input JSON too structurally broken to recover (e.g. completely garbled text) — may return null object
- Schema validation failure if optional `esquema` provided and repaired JSON doesn't conform
- Empty or whitespace-only `texto` input causing parse error
- Network or payment processing errors returning HTTP 402/500

## How this service works

Repair broken JSON, typical of LLM output: strips ```json fences, comments and surrounding text, trailing commas, single quotes, unquoted keys, Python True/None and unclosed brackets. Returns the object and clean JSON and, if you pass a JSON Schema, whether it validates and why not.

## Output

Returns an object containing: the repaired JSON as a string (`json`), the parsed JavaScript object (`objeto`), a boolean `reparado` indicating whether repairs were made, and a boolean `ya_era_valido` indicating whether the input was already valid JSON requiring no changes.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "texto": {
   "type": "string"
  },
  "esquema": {
   "type": "object",
   "description": "optional JSON Schema"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/vextorium-json-repair-521be37e/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from api.vextorium.com](https://www.zero.xyz/host/api.vextorium.com/llms.txt)