Jev Agent

A Jev-first agent loop for DeepSeek Harness: Jev reads the same assembled session and tool history as the main model and either dispatches a saved Action, asks the main model, clarifies, or ends the turn. Requests covered by verified Actions can finish with zero main-model calls.

A developer-preview personal agent harness that turns a sequence of model calls into continuing work: understand the request, choose an action, execute, inspect evidence and recover context. You pick the general model and can optionally connect a Reflex/Jev decision service.

An unofficial Claude Code plugin that checks 'done' claims and small judgement calls with TypeSafe Jev and keeps receipts: its done-gate can reject 'all tests pass' and ask for a real check, so evidence beats eloquence.

An unofficial TypeSafe Jev integration for Claude Code with hooks that suggest skills, classify task type, pick subagent model tiers and gate risky shell commands. It also ships jev-vision for yes/no, pick-one and bounding-box questions about images.

A benchmark asking how well individual decisions add up to a completed task: decision and chat models steer an agent through 100 grid puzzles, where Jev chooses an optimal move in about 90% of fixed exam positions yet finishes only 57 of 100 games.

A Mac desktop bookkeeping clerk where Jev makes every in-app decision — an action kind from a closed list of 12 and which line to click — while a System 2 model rewrites its playbook every ten invoices.

TypeSafe Jev plays NetHack 5.0 with no LLM in the loop: each turn the code lists the legal moves, Jev picks one as a choice question, and a small motor translates the answer into keystrokes, with a live dashboard and Hardfought support.

Documentation hub for wiring Jev into Hermes Agent, Nous Research's autonomous agent framework: Choice, Noul and Score arrive as MCP-style tools, with context curation, turn admission and receipt-backed verification described end to end.

An experiment where Jev plays Zork I: every turn Jericho supplies the valid actions and Jev answers one Choice of which command to type, with confidence bars, an anti-loop memory of the last 8 turns, a JSONL log and a French replay dashboard.

An agent-first CLI for System One: ask, rank, extract, verify, screen, find, and decide from the shell, with an installable agent skill. Published on PyPI.

Drop-in skill for Claude Code, Codex, and other agent environments that teaches the right way to structure states and questions.