What is Pi-JEV-VCC memory relevance?

An extension for the pi coding agent that scores indexed memories against the agent's current context and returns a ranked hint about where useful context may live, using a yes/no classifier read from native token logits.

Scores are a calibrated P(Yes) read from native token logits rather than embeddings. Hosted JEV is the default backend with a zero-dependency install, and a fully local scorer is available for zero-egress operation.

Does it retrieve or inject memories automatically?

No. It does not retrieve or inject memories, and the author calls the scores a prior, not a verdict; the agent alone decides what to read based on the ranked hint.

What can the scorer see?

Only the memory IDs and descriptions in the index. The scorer never reads your memory files, which bounds both what it can leak and what it can reason about.

When does the relevance scoring run?

On three triggers: cold start, after compaction, and from an on-demand tool. That design targets long-running agents that accumulate memory and need to know which parts might matter right now.

What does it cost or require?

The default hosted backend needs no local dependencies but implies a JEV endpoint; the local scorer option exists for zero-egress setups. The captured README does not price calls or document key setup.

Who built it and how mature is it?

It is DAXZEIT's MIT-licensed Python repository, created 2026-10-02 and last pushed 2026-10-13 with no stars or forks; no thresholds or benchmarks are published in the captured text.

Jev directory

CommunityJev Integrations & Adapters

Pi-JEV-VCC Memory Relevance

A pi coding-agent extension that ranks long-term memory files against the current context with a yes/no classifier read from native token logits. It triggers on cold start, post-compaction and an on-demand tool, and offers a prior, not a verdict.

See highlights Open github.com

Stars: 0
Language: Python
License: MIT
Reading time: 2 min
Published by: Community
Author: DAXZEIT
Last verified: Oct 2026

DAXZEIT/Pi-JEV-VCC-memory-relevance

User repository on GitHub

View on GitHub

Logit-scored memory relevance for the pi coding agent: calibrated P(Yes) ranking of long-term memory files on 3 triggers (cold start, post-compaction, on-demand tool). Prior, not verdict.

Forks: 0
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity github python memory logits

Works forJev for Ranking Jev for Scoring

Used inJev for Software & DevTools

Languages: Python
Capabilities: Open sourceRuns locallyRequires API key

Highlights

Scores come from native token logits as a calibrated P(Yes) ranking, not from embeddings.
Runs on three triggers: cold start, post-compaction, and an on-demand tool.
It never retrieves or injects memories; the scores are a prior and the agent alone decides what to read.
The scorer never reads memory files, seeing only memory IDs and descriptions in the index.
Hosted JEV is the default backend for a zero-dependency install; a fully local scorer is available for zero-egress operation.

Watch out

README reviewed 2026-10-17; the captured README stops mid-diagram, the MIT-licensed repository publishes no thresholds, latency or accuracy numbers, and it is explicit that scores are a prior rather than a decision.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

Hacker News: jevmem keeps Claude Code project memory by asking Jev which turns are worth saving.

avinashjetwani61 pointsJevmem – automatic project memory for Claude Code, built on JevRead the thread on Hacker News

From the Jev guides

Original write-ups on what this entry is for.

Decision gates in agent frameworks: what each integration ships

Oct 16, 2026

How Pydantic AI, Mastra, TanStack AI, DSPy, Spring AI, and LiteLLM expose Jev: where the typed decision sits in the loop, what each integration returns, and which big frameworks are only adjacent.

Jev Integrations & Adapters#agents#frameworks#integrations

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai