Jev directory
CommunityJev Agents & MCP
Jev Claud
Claude Code plugin and MCP toolset that compacts test and build output, processes data files outside the context window, and routes decisions to Jev, an LLM, or plain code. Zero cost by default.
- Stars
- 1
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Installs with a one-line PowerShell or shell script and is published on PyPI as jev-claud.
- A hook in ~/.claude/settings.json compacts pytest, npm test, and cargo build output, cutting 70% to 85%.
- MCP tools such as classify, rank, and summarize_file are available to the main agent.
- Ships the Jevin plugin with the jev skill and the scout subagent for code exploration.
- Local summarization runs on Ollama with six effort levels and a JEV_CLAUD_OFFLINE mode; jev-claud spend reports savings.
Questions about this entry
Short answers drawn from the linked project or page.
Run the one-line installer for your platform: the PowerShell script on Windows, or the shell script on macOS and Linux. The assistant asks before each step, then you add the Jevin plugin from Personalization › Skills, or let the assistant install it in Claude Code.
It compacts test and build output before it reaches Claude, cutting 70% to 85% of output for commands like pytest, npm test, and cargo build, and it processes data files outside the context window instead of reading them in full.
The MCP tools include classify, rank, and summarize_file. They are registered at user scope in Claude Code, and a global rule in ~/.claude/CLAUDE.md tells Claude when to use each tool and when to answer directly.
Local summarization runs on Ollama on your own machine, free and without internet, with six effort levels from baixo to ultracode. Jev decisions are not billed, and the default configuration has zero cost.
Only the plugin does. Cloud sessions run on an Anthropic machine that clones the repository, so the hook, MCP tools, global rule, and local Ollama are absent. Use Remote Control to keep the session running on your own computer.
Run jev-claud setup --check to confirm everything is active, and jev-claud update to refresh the package, the Jev SDK, the Jevin plugin, and the global rule. Open new conversations afterwards.
The README and documentation are written in Portuguese. The project is maintained by thiago-lage and distributed as jev-claud on GitHub and PyPI.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: jev-review: a local-first score loop for coding agents
Reddit: One command to wire Jev into Claude Code, Claude Desktop, and Codex.
YouTube: Ray Amjad puts Jev inside an agentic coding loop and looks at what it costs to run.
YouTube: RoboNuggets puts Jev to work inside Claude Code.
More Jev resources like this
Djaevin, a calibrated, local-first decision layer from Icaro0310 that runs behind a Devin ACP backend with no API key. It ships as an MCP server published in the official MCP Registry and installs from PyPI as poordjaevin; the repo topics name calibration, conformal prediction and NLI.
Jev Agents & MCP#community#github#python
Community
icaro0310/devin-judgePython
icaro0310/devin-judgePythonA local System One decision layer for AI coding agents, shipped as an MCP server and Agent Skill, that offloads high-volume classification, checks, scores and ranking from frontier LLMs to sub-50ms local models with calibrated output.
Jev Agents & MCP#community#github#mcp
Community
aktonay/snapdecPython
aktonay/snapdecPythonFive website-building workflows packaged as a Claude Code and Codex skill, standalone Python scripts, and a Node module: page mapping, content migration, silo and doorway checks, and an on-site question router, each with measured runs and costs.
Jev Agents & MCP#claude-code#coding-agent#agents
Communityjackson7705/jev-web-skills
A single-binary MCP tool named evaluate that sends Noul, Choice, and Score questions to hosted Jev or a local System One server.
Jev Agents & MCP#community#go#mcp
Community
itsmostafa/system-one-connectorGo
itsmostafa/system-one-connectorGoReusable project context and specialist workflows for the AI tools you already use: agents, skills, rules, and MCP tools for Claude Code, Codex, Copilot, and Cursor, plus Jev decision support that offloads bounded LLM decisions to save tokens.
Jev Agents & MCP#community#github#agents
Community
jmagly/aiwgTypeScript
jmagly/aiwgTypeScriptAn MCP server and CLI web crawler for AI agents in which a small decision model picks every link and passage, so the agent's LLM reads only what answers its question. On 50 WebWalkerQA tasks the README claims 40% correct answers at about $0.025 per task.
Jev Agents & MCP#community#github#python
Community
rayenmlayeh/jevcrawlPython
rayenmlayeh/jevcrawlPythonFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.