Jsonbit releases JSON compression tool to reduce token consumption when AI reads logs
Understand this story
Jsonbit has released a tool that first compresses nested JSON before passing it to an LLM, aiming to reduce token consumption when AI reads logs. Users can install it via pip install jsonbit and use jsonbit digest to generate compact summaries. The tool also includes a built-in stdio JSON-RPC 2.0 MCP server (jsonbit mcp) that does not depend on an external runtime, allowing agents to query raw data locally with SQL. When installing precompiled native binaries, Python or Node.js users do not need a Rust compiler; building from source requires Rust 1.75+.
AI overview based on source reports · Updated 1 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIJsonbit – Reduce token usage when AI reads logs
Jsonbit is a tool that compacts nested JSON into a 3NF columnar store before it reaches an LLM, offering a one-prompt "digest" of all rows or local read-only SQL/MCP querying so agents read only needed results. It installs via pip, cargo, npm/npx or pre-built binaries, and ships an MCP server exposing tools such as olap_load, olap_digest, olap_ask, olap_sql and olap_rehydrate. The author reports that on 27 questions across 10 workloads, single-prompt digest used 57.4% fewer estimated tokens than minified JSON with 26/27 vs. 13/27 accuracy, and that schema card plus SQL reached 27/27 at 11,729 tokens.
Activity for this story
- Current comparable activity
- 5
- Peak comparable activity
- 10Nov 21 07:00
- Change in the past 24 hours
- -50%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.