How does this stop experiment context from quietly rotting?

Use the per-session loop of start, frame, plan, record, and end, with synthesize promoting traceable observations and correct freezing or superseding bad records. Doctor and claim-audit stay read-only so audits cannot rewrite history.

Why use a decision model instead of letting the main model audit?

One batched call covers all records and questions with a few hundred input tokens in the quickstart triage, returning probabilities and confidence that split findings into silent pass, warning, or manual review within a closed vocabulary.

How do I enable the Jev triage and doctor scripts?

Set TYPESAFE_API_KEY directly or via a git-ignored .env that jev_client.py loads from the working, scripts, or skill-root directory. Real environment variables take precedence and scripts exit 0 with a fallback note on failure.

Will this work with agents other than Claude Code?

Install with npx skills add for project or global scope, or copy the folder into the agent skills directory, then ask the agent to initialize research memory. Daily use is continue, record results, and close the session.

What are the ten operations?

init, start, frame, plan, record, end, correct, synthesize, claim-audit, and doctor. start plus the two audits are read-only; the rest write, and frame writes only after confirmation.

What do I need installed to run the scripts?

Python 3.10 or newer and nothing else: the scripts have zero third-party dependencies. Install the skill with npx skills add poiuyjie/jev_project_context, or clone it into your agent's skills directory.

What is the memory layout?

AGENTS.md or CLAUDE.md as the operational entry, CURRENT.md for present state, EXPERIMENTS.md as the registry, and docs/ with protocols, research, knowledge, experiments, and journal subdirectories. Each file's schema is documented in references/schemas.md.

CommunityJev Practices & Patterns

Jev Project Context

Research-memory skill that keeps experiment claims traceable with ten operations, provenance gates, and optional batched Jev audits over memory files.

See highlights

Open github.com

Stars: 9
Language: Python
License: MIT
Published by: Community
Author: poiuyjie
Last verified: Oct 2026

poiuyjie/jev_project_context

User repository on GitHub

View on GitHub

Evidence-first long-term experiment memory skill for AI coding agents, with optional Jev decision-model layers

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

Evidence-first, long-term experiment memory for AI coding agents. Every claim stays traceable from question to evidence, every session resumable without archaeology.

Quoted from the project's README. Read the full README.

Tagscommunity awesomejev catalog-import

Works forJev for Classification Jev for Tool calling

Used inJev for Software & DevTools

Languages: Python

Highlights

Ten-operation loop from init and task-conditioned start through record, end, and synthesize.
Jev context triage prints LOAD/SKIP manifests while still surfacing stale validity warnings.
Jev doctor pre-screens provenance, headline-table match, and claim support classes.
Zero third-party Python dependencies with schemas, memory layout, and fallback workflows.

Watch out

Jev findings are advisory triage only and never mutate memory; without a key the same checks fall back to the main model with higher cost and latency. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jevmory

Local memory layer that stores verbatim agent quotes in per-project SQLite and grades them with Jev into stale, wrong, unsupported, or keep verdicts.

Jev Practices & Patterns#community#awesomejev#catalog-import

Community

romiluz13/jevmoryPython

laya-browser-agent

Local harness that runs the open Laya decision model on your own machine and wires it into browser agents through a TypeSafe-compatible /v1/systemone dialect.

Jev Practices & Patterns#community#awesomejev#catalog-import

Community

ChenneyZhuang/laya-browser-agentPython

daf-jev

Python client and decision toolkit for the Jev System One endpoint with noul, choice, and score primitives, batched asks, composition helpers, concurrent evaluation, and an MCP server.

Jev Practices & Patterns#community#jevbest#catalog-import

Community

docxology/daf-jevPython

jevloop

Independent open-source agent runtime where Jev makes typed tool decisions, an LLM authors text only when needed and reviews low-confidence calls, and code owns execution.

Jev Practices & Patterns#community#awesomejev#catalog-import

Community

parkavenue9639/jevloopPython

jev-storyboard-lab

Companion code comparing Google ADK and Microsoft Agent Framework on one Pydantic VideoTimeline schema, with the same Jev client checking every generated segment for sane captions and durations.

Jev Practices & Patterns#community#awesomejev#catalog-import

Community

jimmyliao/jev-storyboard-labPython

jet

Coding agent that pairs explicit SQLite-typed context chunks with a small TypeSafe judgment model, so the large generation model only sees selected context.

Jev Practices & Patterns#community#awesomejev#catalog-import

Community

arczhi/jetPython

From the Jev guides

Original write-ups on what this entry is for.

Designing the State: What to Include and What to Drop

Oct 12, 2026

Most Jev accuracy problems are state problems. What belongs in a state, what is a pre-made judgment to leave out, and why naming the slice you ask about matters more than the wording.

Jev Practices & Patterns#state#design#primitives

Read article

Noul, End to End: One Probability and a Dozen Thresholds

Oct 12, 2026

A Noul returns one number and no separate confidence. How to read the value, pin its boundary with criteria, split compound questions, and set a per-action threshold that matches the cost of being wrong.

Jev Practices & Patterns#noul#primitives#confidence

Read article

Scoring Without Measuring: Bands, Rubrics, and Composite Triage

Oct 12, 2026

A Score is a weighted position, not a ruler. Why the number can land between levels, why levels are judged separately, and how to combine several atomic scores in code instead of one crowded rubric.

Jev Practices & Patterns#score#primitives#scoring

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai