pi-jev-context
Opt-in, reversible Pi context filter where Jev scores older messages and low scorers are hidden from requests, not deleted.
Overview
Jev Context for Pi uses TypeSafe Jev to judge which older messages are still useful. Low-scoring content is hidden from future model requests but not deleted from the session, and /jev off restores everything. Pruning is off by default and must be enabled with /jev on after logging in with a TypeSafe key.
User requirements, system instructions, todo operations, extension messages, images, incomplete tool exchanges, summaries and the latest five tool pairs during active work are kept intact. Eligible candidates are complete text-only tool pairs, assistant text and readable thinking, and user shell history in context. A candidate stays only if its keep probability is greater than the threshold, default 0.8, which the README calls aggressive. Commands adjust threshold, buffer and caching, and a branch-aware cache means only new or changed candidates are judged. The README warns that candidate history, tool arguments and results, which may contain private code or secrets, are sent to TypeSafe when enabled.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is this older message or tool pair still useful? | keep · hide |
How it works
- Enable with /login typesafe and /jev on.
- Jev scores eligible older history.
- Hide candidates at or below the threshold from model requests.
- Adjust with /jev threshold or restore with /jev off.
Strengths
- Reversible: session history is untouched.
- Off by default with clear data warnings.
- Threshold changes reuse cached scores.
Caveats
- Default threshold of 0.8 is aggressive.
- Sends history excerpts to TypeSafe when enabled.
Getting started
pi install npm:pi-jev-context, then /login typesafe and /jev on.At a glance
- Type
- Plugin
- Status
- beta
- Language
- TypeScript
- Runtime
- Pi extension
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A reversible Pi context filter using Jev to judge whether older messages remain useful.
logicrw/awesome-jev-projectsOpt-in Pi extension that filters older messages from model requests while preserving the original session history.
cobanov/awesome-jevReversible context pruning for Pi, powered by TypeSafe Jev. Keep useful context without deleting session history.
hellogumbo/awesome-jev
Frequently asked questions
Is anything deleted?
No, low-scoring content is hidden from requests, not deleted.
What is the default threshold?
0.8; lower it to retain more.
Is it on by default?
No, you enable it with /jev on.
Related projects
Codex plugin that builds handoff packets by asking Jev whether each old read-only tool record should stay, keeping retained text verbatim.
ProjectbetaJavaScriptAgent plugin that asks Jev whether the session is at a safe boundary to /compact and can trigger it on Pi and Claude Code.
PluginbetaTypeScriptPi extension where Jev judges each completed tool call and result before compaction, keeping key evidence verbatim in the summary.
PluginbetaTypeScriptOh My Pi extension that truncates tool calls and results Jev scores as no longer needed, while keeping all user and assistant text verbatim.
PluginbetaTypeScriptPi extension that prunes stale tool calls with two Jev yes/no questions each, keeping text verbatim and leaving summaries to Pi.
PluginbetaTypeScriptPi port of fast-jev-compaction using Pi's context hook, with a savings gate against low-value pruning and retained status notes.
PluginbetaTypeScript