jevagentharness/jevagentharness
Organization repository on GitHub
- Forks
- 2
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Agents & MCP
JevAgentHarness
A harness in which a small decision model watches every step of your coding agent: routine steps go to a cheap model, hard steps to the strong one, aiming for the same results at a fraction of the bill. Python 3.10+, MIT, alpha, with an arXiv paper and Hugging Face models under the same name.
- Stars
- 9
- Language
- Python
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- jevagentharness
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open source
Highlights
- A small decision model watches every step of a coding agent and splits steps between a cheap and a strong model.
- Stated pitch: same results, a fraction of the bill.
- Status alpha; Python 3.10+; MIT licence; created 2026-10-15 and pushed 2026-10-16.
- Comes with an arXiv paper (2610.21412) and Hugging Face models and data under the JevAgentHarness name.
- README sections: Results, Quickstart, Connect Your Harness, How Routing Works, and Decision Points and Modes.
Watch out
README reviewed 2026-10-17; the digest stops inside the link row below the badges, so the results numbers, quickstart commands, harness integrations and paper claims could not be read. The project labels itself alpha, and no reproduction of its savings claim is captured here.
Questions about this entry
Short answers drawn from the linked project or page.
A harness that watches every step of your coding agent and routes work: routine steps go to a cheap model, hard steps to the strong one, with the stated pitch of the same results at a fraction of the bill.
It is the component that classifies each step of the agent's work — the README calls it a small decision model — and its routing verdict decides whether the cheap or the strong model handles the step.
Badges link an arXiv paper, 2610.21412, and its Hugging Face page, and the repository publishes models and data on Hugging Face under the JevAgentHarness organisation. The digest does not capture the paper's results.
The README has a 'Connect Your Harness' section and a 'Decision Points and Modes' section, so integrations are configurable, but the digest stops before those sections, so the actual steps and supported harnesses are not captured here.
The premise is that a cheap model handles routine steps and the strong model only the hard ones, so the bill should shrink — but the digest shows no measured cost or savings figure, only the project's own 'fraction of the bill' pitch.
No — the badge row labels it alpha, alongside Python 3.10+ and an MIT licence. Treat it as a research-grade release; the repository was created 2026-10-15 and pushed 2026-10-16.
It lives under the JevAgentHarness organisation on GitHub, with a project page at jevagentharness.github.io and artifacts on Hugging Face. Metadata: Python, 5 stars, created 2026-10-15, pushed 2026-10-16.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.
FacebookClinicnotes
X: OpenRouter Decision Model Rankings
X: herdr + jev, an agent router
X: A group chat bot with decision room routing
More Jev resources like this
A free Claude Code skill that puts Jev in front of the expensive model: thirteen commands sort, check, score, gate and verify with Jev, so Claude reads only the unsure slice and writes the output. Needs python3 and an OpenRouter key.
Jev Agents & MCP#community#python#claude-code
Community
charlesdove977/claude-x-jevPython
charlesdove977/claude-x-jevPythonClaude Code hooks that narrow a roster in YAML, ask Jev which handler owns a prompt, and log the band before anything is blocked.
Jev Agents & MCP#community#python#coding-agent
Community
AgriciDaniel/gatekeeperPython
AgriciDaniel/gatekeeperPythonAn event-driven multi-agent network where JEV observes live state and assigns pending work to available agents. Its AgentRouter reference implementation routes on every completion event instead of waiting for a round to end, while general-purpose agents with OpenAI Responses API tools execute work concurrently.
Jev Agents & MCP#community#github#agents
Community
Kelly-Leeee/AgentNetwork-jevPython
Kelly-Leeee/AgentNetwork-jevPythonAn agent harness where Pi runs the loop, Gemini does the work, and Jev makes the small decisions: which model tier to use, which runbook sections to load, whether a tool call may run on a real Postgres database, and whether the report holds up.
Jev Agents & MCP#community#python#agents
Community
PromtEngineer/jev-harnessPython
PromtEngineer/jev-harnessPythonA Hermes plugin that routes every message you type to one of four model levels with Jev deciding in about 300 ms: lightweight for the trivial, maximum for what matters, and continuations stay on the model with the warm cache.
Jev Agents & MCP#community#python#hermes
Community
okjpg/jev-hermes-routerPython
okjpg/jev-hermes-routerPythonDjaevin, 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-judgePythonFrom 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.