Hermes Jev Skills

Give your agent a fast, cheap second brain for the small decisions.

Jev is TypeSafe's decision model. It does not write text. You hand it a state and typed questions (pick one, score this, yes or no) and it answers in about 0.4 seconds for a tiny fraction of a cent, with a calibrated confidence. This repo puts that to work on the decisions an agent makes all day, so your expensive model only does the thinking and writing.

Plus a model routing dashboard (jev dashboard): every profile's models on one page, an on/shadow/off switch for Jev routing, and a live view of where each turn is being sent. See router-dashboard.

Built for Hermes. The skills and the jev command also work in Claude Code, Codex and anything else that reads SKILL.md files.

Install

Point your agent at this repo and say: "Install Hermes Jev Skills from https://github.com/kerpopule/hermes-jev-skills". It will follow AGENTS.md.

Got it as a zip? Unzip it anywhere, then run the second and third commands below from that folder.

Or by hand (Python 3.9+, no dependencies):

git clone https://github.com/kerpopule/hermes-jev-skills ~/hermes-jev-skills

python3 ~/hermes-jev-skills/install.py

jev setup-key

The installer finds Hermes, Claude Code and Codex on the machine and installs for each one it finds. python3 install.py --check shows what it would do without changing anything; --uninstall reverses it.

Your API key never touches the agent

jev setup-key opens a one-time page served only by your own computer. You paste your TypeSafe key there. It goes straight into the OS secret store (macOS Keychain, or secret-tool on Linux, or a 0600 file as a last resort) and, on a Hermes machine, into each profile's .env. The agent that ran the command sees one line: stored, verified, yes or no. Never the key, not even a prefix.

The page lives on an unguessable one-time URL, refuses requests with a foreign Host header (DNS rebinding), sends no referrer, logs nothing, and shuts down after one use or ten minutes. On a headless box, run jev setup-key --tty yourself for a hidden prompt.

Do not paste your key into a chat. If you already did, make a new one.

On Hermes

The hermes-jev plugin uses only public plugin seams (pre_llm_call, llm_request middleware, tools, a slash command), so hermes update does not break it and nothing in Hermes core is patched.

/jev                         status
/jev routing shadow          decide and log, do not switch (start here)
/jev routing on              switch models per turn
/jev skills on               suggest the right skill per turn
/jev notice on               show "[Jev] hard · coding → kimi-k3 · confidence 0.92" on routed replies
/jev routing on all          make it the default for every profile (a profile's own setting still wins)

Tools the agent gets: jev_memory_filter, jev_compact_select, jev_choose_action.

A plugin can swap the model, not the provider connection. On OpenRouter that still covers every vendor. If you run /model yourself, your choice wins.

Every model you have

jev models list              # everything this machine can call: price, context, vision, reasoning
jev models providers         # which providers you hold a key or login for
jev models suggest --write   # first-draft pools from price bands; then edit to taste

The catalog is 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. Pools live in ~/.hermes/jev/routing.json (the default for every profile); ~/.hermes/profiles/<name>/jev/routing.json overrides it for one profile. See skills/jev-model-routing.

What leaves your machine

Jev is a cloud API, so this is spelled out:

• Routing: the user's turn, redacted (emails, phones, tokens, long hex masked), capped at 3,000 characters. Never history, tool results, files or memory. Turns that look like they hold a secret, and any profile you list in private_profiles, send only coarse features: length, whether code is present, whether risk words appear.
• Memory: the query and up to 900 characters per passage, redacted. Your store's ids, paths and sources are replaced with P0, P1… and never sent. A passage that looks like a credential is not sent at all.
• Compaction: up to 700 characters per turn, redacted. Turns that look sensitive are skipped.
• Skills: the turn, redacted, plus skill names and descriptions.
• Computer and browser use: the goal, short element labels, and your action descriptions. Never screenshots, page text or field values. A goal or label that looks sensitive is refused before sending.

Logs hold decisions only (tier, model, confidence, latency). Never prompt text.

Everything fails open

No key, timeout, rate limit, malformed reply, low confidence: routing keeps your current model, memory returns the original list, compaction drops nothing, skill selection suggests nothing, and computer use returns reobserve. A Jev outage costs you at most the time budget (2.5 s for routing) and never blocks a turn.

Safety rails that do not depend on Jev being right: risk words (production, delete, migration, security, payment, legal…) never route to the cheapest tier; a large context never switches to a cheaper model mid-session; a transcript turn is only dropped on a confident answer; Jev can only ever return an action id you put in the table.

Layout

jevkit/          the library and the `jev` command (stdlib only)
skills/          eight SKILL.md skills, agent-agnostic
hermes/plugin/   the Hermes plugin
router-dashboard/  the model routing page (`jev dashboard`)
install.py       installer / uninstaller
tests/           offline tests, every Jev reply faked
docs/            integration notes and hard-won operational lessons

python3 -m unittest discover -s tests

License

MIT. Jev and TypeSafe are products of TypeSafe AI; this project is independent. Optional browser runner wraps browser-use/jev-ultrafast (MIT), which is not bundled.