User repository on GitHub
Jev-powered model routing, memory, compaction, skill selection, computer and browser use for Hermes agents (also Claude Code and Codex)
- Forks
- 115
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Hermes Jev Skills
A Python toolkit that installs nine plain SKILL.md files for Hermes, Claude Code, and Codex, letting Jev route models, filter retrieved passages, select skills, and choose bounded computer or browser actions.
- Stars
- 1.1k
- Language
- Python
- License
- MIT
- Reading time
- 4 min
- Published by
- Community
- Author
- kerpopule
- Last verified
- Sep 2026
About this project
Give your agent a fast, cheap second brain for the small decisions. Your agent burns frontier-model tokens on things that are not thinking: which model should answer this turn, which of 377 skills to load, which retrieved passages matter, which turns survive a summary, which button comes next. Those are decisions, not prose. Hand them to something that costs a fraction of a cent and answers in about 0.4 seconds, and let the expensive model do the writing.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
Highlights
- Nine skills ship as plain SKILL.md files that work in Hermes, Claude Code, Codex, or anything that reads a skill file, with Python 3.9+ and no dependencies.
- Measured costs in the README: about 0.4 s per routed turn, 2.8 s to choose among 377 skills, and $0.00006 per triage message.
- Routing starts in a shadow mode that decides and logs without switching anything, and install.py --check shows what would change before it does.
- Every skill documents a safe default when Jev fails or is unsure: routing keeps the current model and memory returns the original list.
- The API key stays in the OS keychain rather than reaching the agent, and the test suite is offline with every Jev reply faked.
Quickstart
bash
git clone https://github.com/kerpopule/hermes-jev-skills ~/hermes-jev-skills
python3 ~/hermes-jev-skills/install.py
jev setup-key
jev doctorWatch out
MIT-licensed and independent of TypeSafe. Needs Python 3.9+ and a TypeSafe key; enabled features send redacted prompts or excerpts to TypeSafe, and suggest and calibrate read local session transcripts.
Questions about this entry
Short answers drawn from the linked project or page.
No. Routing sends a redacted turn capped at 2,500 characters and never history, tool results, files or memory; memory filtering sends the query plus up to 900 characters per passage with your store's ids replaced by P0, P1. Profiles in private_profiles send only coarse features like length and whether risk words appear.
The scorecard itself flags it as not portable recall. Across 80 real web results with in-distribution planted attacks it caught 70 of 79 against 11 for Hermes's own pattern scan, and withheld 0 of 1,520 clean chunks. Read it as a filter that adds signal, not a security boundary you can rely on alone.
It can, and on handoffs the repo reports the negative result: a digest written from Jev's keep, summarize or drop labels recalled less than one written from the plain transcript, so what ships is the whole dialogue plus a way back to the old session. Turn selection does beat recency, 71 turns in 0.95 s and 11 questions to 4.
Not if you follow the documented path. Routing starts in shadow mode, deciding and logging without switching anything, and install.py --check prints exactly what would change before touching anything. Even when on, a plugin can swap the model but not the provider connection, and running the model command yourself wins.
Model routing, search, social research, memory, web screening, choosing turns, skill selection, triage, and computer and browser use. Slash commands cover status, routing shadow or on, skills on, notice, and screen shadow or on, and the agent is handed five tools: jev_memory_filter, jev_compact_select, jev_choose_action, jev_supervise, and jev_escalate.
From models.dev, filtered to providers whose API-key name is set in your environment or Hermes .env, or that Hermes holds a login for. Only key names are read, never key values. Pools live in ~/.hermes/jev/routing.json as the default for every profile, with ~/.hermes/profiles/<name>/jev/routing.json overriding it for one profile, and jev models suggest --write drafts a first pass from price bands.
jev setup-key serves a one-time page bound to your own computer, on an unguessable URL that refuses requests with a foreign Host header to block DNS rebinding, sends no referrer, logs nothing, and shuts down after one use or ten minutes. The key goes straight into the OS secret store and, on a Hermes machine, each profile's .env. The agent that ran the command sees one line: stored, verified, yes or no.
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 cross-platform memory layer for AI coding agents: conversations from Claude Code, Codex, DeepSeek Harness, OpenClaw, and OpenCode become Markdown memories indexed in Milvus, with optional Jev reranking through the TypeSafe API.
Jev Tools & CLIs#community#python#agents
Community
zilliztech/memsearchPython
zilliztech/memsearchPythonA collection of Jev demos, workflows, and agent skills for coding agents: five installable skills and 108 scenarios, installed by pointing Codex, Claude Code, or OpenCode at an agent prompt.
Jev Tools & CLIs#community#python#skills
Community
wuyoscar/jev-skillPython
wuyoscar/jev-skillPythonA reference router and skill that put Jev in front of Grok Bot so each request is classified before expensive research, browser, retry, or subagent work, returning actions such as reuse_cache or ask_human.
Jev Tools & CLIs#community#python#routing
Community
Bodila51/grok-bot-jevPython
Bodila51/grok-bot-jevPythonA local-first Herdr CLI that filters your agent subscriptions with fixed eligibility rules, then asks TypeSafe Jev to rank the remaining Cursor, Claude Code, Codex, or OpenCode options and pick a reasoning effort.
Jev Tools & CLIs#community#typescript#cli
Community
nidhi-singh02/agent-routerTypeScript
nidhi-singh02/agent-routerTypeScriptLangChain's langchain-typesafe integration exposes Jev as a TypeSafeClassifier plus experimental model-routing and auto-mode middleware that screens risky tool calls before execution.
Jev Tools & CLIsArticle#community#integration#python
Community
A Go-based team workspace with optional Jev AI for agent auto-routing, task auto-fill, auto-assign, and automation condition nodes.
Jev Tools & CLIs#community#go#agents
Community
Paca-AI/pacaGo
Paca-AI/pacaGoFrom 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.