Repair malformed LLM JSON API

Fixes broken JSON an LLM emitted — strips markdown fences and prose, repairs trailing commas, single quotes, unquoted keys, Python literals (True/None), and closes truncated strings/brackets — returning valid JSON plus the parsed value and the list of fixes applied. Answers 'how do I parse this model output that isn't quite valid JSON?', 'how do I recover a truncated JSON response?'.

Price$0.05per request

MethodPOST

Route/v1/llm/json-repair

StatusLive

MIME typeapplication/json

Rate limit120/minute

Cache0s public

llmjsonrepairparsestructured-outputfixtruncatedagent

API URLhttps://x402.hexl.dev/v1/llm/json-repair

Integration docs

Example request

{
  "text": "```json\nHere: {name: \"Bob\", age: \"30\", tags: [1,2,],}"
}

Example response

{
  "ok": true,
  "repaired": "{\"name\": \"Bob\", \"age\": \"30\", \"tags\": [1,2]}",
  "value": {
    "name": "Bob",
    "age": "30",
    "tags": [
      1,
      2
    ]
  },
  "changed": true,
  "fixes": [
    "dropped-leading-prose",
    "removed-trailing-commas",
    "quoted-bare-keys"
  ]
}

Input schema

{
  "type": "object",
  "required": [
    "text"
  ],
  "properties": {
    "text": {
      "type": "string",
      "description": "The malformed/truncated JSON or LLM response to repair.",
      "examples": [
        "```json\nHere: {name: \"Bob\", age: \"30\", tags: [1,2,],}"
      ]
    }
  }
}

Output schema

{
  "type": "object",
  "additionalProperties": true
}