Will this run up my bill with a call on every action?

Possibly, because reflexes fire around every tool and stop. The README says it wins on decision-heavy scoped work, every call is metered for stats reporting, and you can disable reflexes or narrow route.tools per project.

What happens when Jev goes down mid-session?

It fails open: disabled, keyless, timed-out, erroring, or below-threshold hooks exit 0 and the loop proceeds untouched. Keyless installs run silent MOCK answers so nothing breaks.

Does Jev overrule Claude on safety or completion?

No. Check and Warden hand Claude a concern to re-examine with the probability shown; Claude still decides. The first action is blocked until Jev is consulted, but that enforces consultation, not obedience.

How do I turn down the chatter for a project?

Drop a .jeveloper.json with drive, route, check, or warden toggles, thresholds like fail_threshold and done_threshold, plus provider selection. Deep reasoning mode stays off unless you enable it.

How do I install jeveloper?

/plugin marketplace add pur4v/jeveloper then /plugin install jeveloper. Install prompts for an OpenRouter API key and stores it in your OS keychain, never in the repo or settings.json, then restart Claude Code once so the hooks register. You can also export OPENROUTER_API_KEY or TYPESAFE_API_KEY.

What are the three reflexes?

Route on PreToolUse decides whether a call is destructive or off-goal and denies or asks; Check on PostToolUse decides whether the output actually succeeded and did its job; Warden on Stop decides whether the task is genuinely complete and verified.

/jeveloper:drive makes Claude enumerate 2 to 5 candidate actions, Jev picks one with a fast choice call, Claude executes it, Jev verifies the output, and Jev judges completion, repeating as a ping-pong. The turn's first action is blocked until Jev is consulted.

CommunityJev Agents & MCP

Jeveloper

Claude Code plugin that puts Jev judgments in the agent loop to route tool calls, verify outputs, judge completion, and pick next actions with metered calls.

See highlights

Open github.com

Stars: 1
Language: Python
License: MIT
Published by: Community
Author: pur4v
Last verified: Oct 2026

pur4v/jeveloper

User repository on GitHub

View on GitHub

A System-1 reflex layer for Claude Code, powered by TypeSafe AI's Jev — route/gate tool calls, verify output, judge completion, and compose decision trees.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

A System-1 reflex layer for Claude Code, powered by Jev. Jev is a System One model from TypeSafe AI — fast, intuitive, one-pass, and typed. It doesn't write prose; it takes state + typed questions and returns typed probabilistic decisions in 70–500 ms for $0.042 per million input tokens (output free). Claude is System Two: deliberate, powerful, expensive.

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

Tagscommunity github

Works forJev for Routing Jev for Tool calling

Used inJev for Software & DevTools

Languages: Python

Highlights

Adds Route, Check, and Warden reflexes on PreToolUse, PostToolUse, and Stop hooks with thresholds.
Blocks the turn first action until Jev is consulted and can drive next-action picks via directives.
Fails open without a key or on errors, running silent MOCK answers so work proceeds untouched.
Reports usage via stats and tunes via .jeveloper.json, with deep mode for hard decisions.

Watch out

Heavy by design with calls around every action; generation itself cannot be offloaded and savings are estimates until benchmarked. 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

agent-harness-with-jev-llm

Demo of the LangChain Jev harness pattern in LangGraph: a fast classifier routes events and gates a risky tool call, leaving open-ended reasoning to a chat LLM. Entry is POST /events; output lands in JSONL files.

Jev Agents & MCP#community#github

Community

dhirajpatra/agent-harness-with-jev-llmPython

langgraph-jev-router

LangGraph conditional edge that uses Jev parallel Choice routing between specialized agents instead of a generative model.

Jev Agents & MCP#community#github

Community

JTankariya/langgraph-jev-routerPython

vexjoy-agent

441

A four-layer agent toolkit for Claude Code, Codex, Factory, and Reasonix: 43 agents, 62 skills, 78 hooks, 147 scripts, and a Jev-backed router that gates dispatch on intent alignment.

Jev Agents & MCP#community#jevbest

Community

notque/vexjoy-agentPython

muse-jev-playbook

Agent-oriented playbook for using Jev as a cheap, fast decision layer that triages, classifies, scores, and gates work before expensive browser, research, retry, or subagent steps.

Jev Agents & MCP#community#awesomejev#catalog-import

Community

Bodila51/muse-jev-playbookPython

Tiermem

Tiered agent memory keeping compact summaries with source links over searchable raw history, where a router decides per query whether notes suffice or raw pages must be recovered.

Jev Agents & MCP#community#jevbest#catalog-import

Community

FreedomIntelligence/TiermemPython

liteagents

Python SDK that runs the same prompt, tools, and MCP setup across DeepAgents, Pydantic AI, Claude Agent SDK, Codex, and OpenCode by switching harness.

Jev Agents & MCP#community#jevbest#catalog-import

Community

BerriAI/liteagentsPython

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai

Jeveloper

About this project

Highlights

Questions about this entry

Will this run up my bill with a call on every action?

What happens when Jev goes down mid-session?

Does Jev overrule Claude on safety or completion?

How do I turn down the chatter for a project?

How do I install jeveloper?

What are the three reflexes?

What is driver mode?

More Jev resources like this

From the Jev community