Features

📐 Depth Limiting

Set a maximum nesting depth
Deeper objects replaced with {"...": "..."}
Deeper arrays replaced with ["..."]
Default depth of 3 levels

🔑 Key & Array Limits

Keep only first N keys per object
Truncate arrays to first N items
Appended count of omitted entries
Set 0 for unlimited keys

📝 String Truncation

Trim long string values
Shows remaining character count
Configurable length threshold
Keeps all other value types intact

JSON Truncation Guide

How Truncation Works

All four truncation modes can be combined. The tool walks the JSON tree and applies each rule at each node:

Max depth — any node deeper than this level is replaced with a placeholder value.
Max keys — only the first N keys of each object are kept; a "..." key shows how many were omitted.
Max array items — only the first N elements of each array are kept; a trailing string item shows the omitted count.
Max string length — string values longer than N characters are sliced and appended with a char-count suffix.

Common Use Cases

API Debugging: Paste a huge REST response and see just the top-level structure before drilling in.
Log Inspection: Trim deeply nested log payloads to find the field you care about quickly.
LLM Prompts: Reduce token usage by sending a representative slice of a large dataset.
Documentation: Generate concise example snippets from real production data.
Schema Discovery: Expose the shape of an unknown JSON document without reading thousands of lines.

FAQ

Q: Is the output valid JSON?

A: Yes. All placeholder values are valid JSON strings or objects, so the output parses cleanly with any standard JSON parser.

Q: What does "0 = all" mean for Max keys?

A: Setting Max keys to 0 disables that constraint entirely — all keys are kept at every level. Only depth, array, and string limits still apply.

Q: Does this modify my original data?

A: No. Processing happens entirely in your browser and the original input is never altered. The truncated copy is written to the output panel only.

Truncating JSON for logs, docs and LLM prompts

Real payloads are often far larger than anyone needs to read. Truncating keeps the structure recognizable while cutting size, which is ideal for bug reports, documentation examples and context passed to AI models.

Tips and common pitfalls

When sharing data with an LLM, keep a couple of array items as examples and drop the rest. The model learns the shape without paying for every record; see our LLM Token Counter to measure the savings.
Shorten long strings such as Base64 images or HTML blobs, which are usually noise when debugging.
Check the truncated output for personal data before sharing it; truncation shortens values but does not remove them.
Keep the original file. Truncated data is for reading, not for processing.

JSON Truncator