Extract Fields from JSON — keep only what matters

About this tool

Input
{ "user": { "name": "Ada", "email": "ada@example.com" }, "debug": true }

Output
{ "name": "Ada", "email": "ada@example.com" }

Extracting fields from JSON keeps an explicit allowlist of paths and drops everything else. It's the inverse of "remove these noisy keys" — instead of describing what to discard, you describe what to keep, and the rest falls away. For deeply nested API responses, log payloads, or LLM context windows, that's the cheaper mental model: list the 5 fields you actually need and ignore the other 200.

The tool walks the document by path and pulls the requested fields into a compact output. Dot-paths address nested values (user.email), bracket indices reach into arrays (items[0].sku), and bulk patterns let you keep all entries that match a prefix. Missing paths are silently skipped — extracting user.phone from a record that doesn't have one yields nothing for that key, not an error, so partial inputs don't crash the pipeline.

This runs entirely in your browser. Paste the JSON, list the paths, and copy the result. No upload, no signup, no rate limit on payload size beyond browser memory.

How to extract fields from JSON

1. Paste your JSON document — an API response, a log line, or anything else — into the input panel.
2. List the paths you want to keep, one per line. Use dot-notation for nesting (user.address.city) and bracket-notation for arrays (items[0]).
3. Click Run. The output is a trimmed copy of the input with everything outside your list removed.

Common use cases

• Trim API responses down to the few fields your code actually consumes.
• Strip PII from logs before sharing with non-engineers — drop email, phone, address, keep the rest.
• Shrink LLM context payloads to fit a token budget. Pair with JSON for LLMs for token-budget tactics.
• Stage trimmed data for export with the JSON to CSV converter — fewer columns, cleaner spreadsheets.
• Flatten the result with Flatten JSON when downstream tools want flat keys (user.name).
• Hand a trimmed payload to the Clean JSON tool for null-stripping and key sorting.

Path syntax notes

Paths address values inside the document:

• name — top-level key.
• user.email — nested key, dot-separated.
• items[0] — first element of an array.
• items[*].sku — every sku across the array (bulk pattern).
• user.address.* — every key under user.address.

Common errors when extracting fields

• Path typos silently drop the field. Missing paths produce no output for that key. Double-check spelling — extracting user.firstname from a doc with user.firstName returns nothing.
• Arrays vs objects. items[0] reaches an array element; items.0 does not. Use bracket notation for arrays.
• Empty result for an entire input. That means none of your paths matched. Verify the input shape with the JSON Diff tool against a known-good record.

Why this tool

• Bulk patterns avoid the "100 paths" mistake when you really mean "everything under metadata".
• Pipeline-native — the output drops straight into the next step without a JSON.parse round-trip.
• Path errors are recoverable, not fatal: missing paths skip silently so partial inputs still produce usable output.

Related tools

• Clean JSON
• Normalize JSON
• Extract Fields from JSON
• JSON Formatter