Dev tools

fast-jev-compaction prunes Claude Code tool calls instead of writing a summary

September 18, 2026 at 1:20 AM PT

The GitHub social card for tamaratran/fast-jev-compaction

Image: GitHub

Why it mattersCompaction rewrites long sessions into a summary that quietly loses file paths, exact errors, and constraints, so a decision-only approach keeps the parts a later step needs.

Tamara Tran published fast-jev-compaction, a Claude Code plugin that changes how the tool trims a long session. Instead of writing a shorter summary of the past, it decides one tool call at a time whether the call and its output are still needed, and keeps everything else exactly as it was. The repository is one day old and has 1,713 GitHub stars, 81 forks and 12 open issues as of this morning. It is MIT-licensed.

How it decides what to drop

For every tool call in the transcript, the plugin sends Jev, a small classification model from TypeSafe, two yes-or-no questions: should the call itself stay, and should its full result stay. Jev sees the whole conversation, oldest first, with each result replaced by a short note about its length. User and assistant text is never rewritten.

Above the keep threshold, which defaults to 0.5, both the call and its result stay. Below it but with a positive vote on the call itself, the call stays and the result is cut to its first 300 characters. Below both, the call is removed with its result. The newest six messages are always pinned and never touched.

What is different from a summary

The built-in Claude Code compaction asks a large model to rewrite the past as prose. That summary is short but lossy: file paths, exact error strings, the one command that worked, or a constraint the user stated once can disappear even when a later step needs them. The plugin's author states that everything kept is kept verbatim, which is the point of the design.

Reduction is not free. Jev sees the entire abridged history with every request, so a session close to the state ceiling can cost one request per small batch of questions. The library exposes reductionRatio, and the README suggests skipping the swap when the ratio comes in under 25 percent, since keeping the original transcript may then be the better option.

The plugin requires Claude Code 2.1.274 or newer and the function-hooks opt-in flag, plus a TypeSafe API key. When Jev cannot prune enough or fails, the plugin falls back to the built-in summary.

For a team running long agent sessions, the tradeoff is concrete: pay for a small classification model on every compaction, keep the raw details the assistant may need again, and skip the guessing about what the summary silently dropped.

Source

tamaratran/fast-jev-compaction on GitHub

Source: GitHub

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

More from AI News

OpenSpec crosses 68,000 stars as a shared spec framework for 40 AI coding assistants, and publishes a v1.13.1 release the same day

OpenSpec is an MIT-licensed spec framework by Fission-AI that works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, Zed and 34 other AI coding assistants, and it reached 68,579 GitHub stars a day after its v1.13.1 release.

Source: Hacker NewsDev tools

ai-data-extractor copies chat history from your machine as JSONL, and reads Claude Code, Cursor, Windsurf, Aider, Cline and eight others

The ai-data-extractor project is an MIT-licensed local tool that searches the disk for chat histories from Claude Code, Cursor, Windsurf, Aider, Cline and eight other coding assistants and normalises them into a single JSONL format, and it has gained 825 stars in five days.

Source: GitHubDev tools

Tokentab reads Claude Code, Codex and Gemini CLI logs and reports the bill by model, project and day

Tokentab is a new MIT-licensed CLI that reads the session logs Claude Code, Codex and Gemini CLI already write to disk and totals the token cost by model, project and day, and the repository has picked up 1,139 stars in nine days.

Source: GitHubDev tools