UNQUOTE

Product updates

What is changing in Unquote

See what changed in Unquote and how the local JSONL viewer is becoming faster, clearer, more reliable, and privacy-first.

Where the product is heading

Unquote expands stringified JSON and turns supported Agent logs into readable sessions without sending imported content to a server. Recent releases focus on maintaining responsiveness as JSONL files grow, explaining Agent activity more clearly, and protecting the exact data users inspect, copy, and export.

Large files, responsive tools
Opening, searching, inspecting, and exporting JSONL should stay usable as files grow.
Agent logs that tell a story
Timelines and trajectories should explain what happened, when, and why it matters.
Correct output before convenient output
Exact numbers, complete records, and explicit limits protect the data being debugged.
Local processing by default
Imported content stays in the browser while the product gains better local analysis.

Latest releases

Release notes written around the change a user can see, not the implementation behind it.

Version 1.2.4

Faster parsing and search, with new web guides

JSON opens with less waiting, submitted searches run immediately, and narrow-screen controls stay accessible. New web guides provide examples for common tasks.

  • Open JSON with fewer unnecessary parsing attempts and no fixed delay before background parsing starts.
  • Submit a toolbar or command search to run it immediately; on-demand views appear as soon as they load.
  • Use search, view controls, and record filters on narrow screens.
  • Explore web guides for JSONL, JSON unescaping, and Agent logs, and open their examples in the viewer.

Version 1.2.3

Compare JSON, inspect fields, and export redacted reports

New local analysis tools help compare values, filter records, inspect field quality, and prepare problem excerpts. Large-file copy and export use less memory and keep the page responsive.

  • Compare pasted JSON, files, or source records by path, ignore selected fields, and see context around changes in long values.
  • Build record tables from nested fields, filter by values or field profiles, and export matching rows as CSV with source line numbers.
  • Select source lines, redact specified paths, preview a problem excerpt, and download Markdown or JSONL.
  • Large-record operations use less memory, failed selection imports offer recovery guidance, and bundled fonts reduce external requests.

Version 1.2.2

Safer export limits, faster agent starts, and unified icons

Heavy exports now protect memory limits with clear feedback, Agent sessions reach their first view with less duplicate parsing, and icons are unified across the interface.

  • Buffered exports and unhydrated copy actions enforce explicit memory limits instead of freezing the tab.
  • Initial Agent session inputs parse once during load, making long logs faster to open.
  • Icons across the app are migrated to Phosphor Icons, and the import panel gains theme-aware backgrounds.

Version 1.2.1

Large JSONL stays responsive from open to export

Heavy local files now use less memory and keep the interface responsive from first open through final export, while stricter data checks reduce the chance of incomplete results.

  • Opening and inspecting large local JSONL files no longer ties up the interface with parsing work.
  • Large exports are produced incrementally, lowering memory spikes and keeping the page usable.
  • Exact numbers, full-source path queries, stable search navigation, and stricter output checks reduce silent data mistakes.

Version 1.2.0

Nested JSON expands all the way through

Consecutive layers of stringified JSON now open recursively, and Agent sessions reach a useful first view sooner with clearer failure details.

  • Objects, arrays, primitives, and strings nested across multiple escaped layers expand consistently.
  • Japanese is available across the web app and browser experiences.
  • Switching sources cannot let unfinished work from the previous source replace the current workspace.

Version 1.1.0

See an Agent session as a trajectory, not just a log

The Trajectory view turns a long Agent session into timed work, connecting turns, tool activity, failures, token usage, and original records.

  • A time overview keeps active clusters readable by compressing long idle gaps.
  • Range, kind, status, search, and failure filters narrow large sessions quickly.
  • Detail views connect calls and results to bounded raw JSON and their full source records.

Version 1.0.1

Safer inspection for large and exact values

Unquote now preserves the exact source text of JSON numbers and states clearly when a preview, copy, or structure measurement is incomplete.

  • Large integers and high-precision decimals survive tree labels, search, copy, and export unchanged.
  • Oversized previews and copies stop at explicit limits instead of freezing or failing silently.
  • Long values remain scrollable, and partial structure facts are no longer presented as complete.

Version 1.0.0

A focused workspace for everyday JSONL debugging

The 1.0 release shaped Unquote around one direct workflow: import a source, find the relevant record, inspect the value, and export the result.

  • A dedicated import flow supports paste, drop, file selection, samples, and format detection.
  • The record rail, selected tree, and node inspector keep navigation and detail visible together.
  • Recognized Agent logs gain a dedicated session view while the original JSONL remains available.