# Grokzilla JSON Repair

> Grokzilla JSON Repair is a paid API for AI agents from grokzilla.shop, paid per call via x402, $0.0041/call, status unknown (last checked 2026-10-18).

Repairs malformed or truncated JSON (from LLMs, JS, or Python) by fixing fences, comments, trailing commas, single quotes, unquoted keys, and unclosed structures, returning the parsed object, a compact repaired string, and a list of applied fixes.

## Facts

- Endpoint: POST https://grokzilla.shop/api/skills/json-repair?utm_source=zero.xyz
- Price: $0.0041/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-18
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/grokzilla-json-repair-7359d20a
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_KhapevPwWcq41zxKWC7Ax

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 grokzilla-json-repair-7359d20a -d '<json body>'
```

Example prompt: The JSON I got back from the LLM is broken — it has backtick fences, trailing commas, and some Python-style True/False values. Can you repair it and give me back the fixed parsed object along with a list of what was changed?

## When to prefer this

Choose this endpoint when you need deterministic, offline-safe JSON repair with an explicit audit trail of changes applied — especially for cleaning LLM outputs that use Python-style literals, markdown fences, or non-standard syntax. Prefer it over a general LLM re-prompting approach when you want reproducible results without network round-trips or hallucination risk, and when you need the specific list of fixes for logging or debugging.

## Known failure modes

- Input is so severely malformed that no valid structure can be inferred — returns an error or empty result
- Input exceeds 400,000 character limit — rejected with a validation error
- Input is valid JSON already — returned as-is with an empty fixes list (not a failure, but callers should be aware)
- Ambiguous repair choices (e.g. multiple interpretations of broken nesting) may produce unexpected but valid JSON

## How this service works

Repair malformed or truncated JSON from LLMs, JS or Python: strips ```json fences and comments, fixes trailing commas, single quotes, unquoted keys, True/False/None, missing commas and unclosed strings/brackets. Returns the parsed JSON, a compact repaired JSON string and the list of fixes. Deterministic, no network.

## Output

Returns three things: the fully parsed JSON object (ready to use programmatically), a compact repaired JSON string, and an ordered list of fixes that were applied (e.g. 'removed ```json fence', 'replaced trailing comma', 'converted True→true'). No network calls are made; processing is fully deterministic and local.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "text": {
   "type": "string",
   "maxLength": 400000,
   "minLength": 1
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/grokzilla-json-repair-7359d20a/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from grokzilla.shop](https://www.zero.xyz/host/grokzilla.shop/llms.txt)