hermes-jev-skills
Skill pack handing an agent's small decisions to Jev: model routing, skill choice, search, memory, compaction and computer use.
Overview
Hermes Jev Skills gives an agent a fast, cheap second brain for small decisions that otherwise burn frontier-model tokens: which model should answer this turn, which of hundreds of skills to load, which retrieved passages matter, which turns survive a summary, which button comes next. Jev never writes text; it answers typed questions with calibrated confidence, and the expensive model keeps the writing.
The README reports about 0.4 seconds per routing turn and about 2.8 seconds to pick among 377 skills, with acknowledgements answered locally for free. Search uses Jev to pick which fetched results to open, whether evidence answers the question and which candidate query to run next, without writing queries; social research and memory (including hidden-instruction checks) follow similar patterns. According to a list summary, a measured handoff digest recalled less than the plain transcript, so handoffs keep the dialogue. A jev dashboard shows routing decisions live. It works with Hermes, Claude Code and Codex, installs with one Python script and has no dependencies.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which model is good enough for this turn? | — |
| Choice | Which installed skill does this turn need, or none? | — |
| Yes / no | Does the fetched evidence answer the question? | yes · no |
| Yes / no | Does this retrieved passage contain hidden instructions? | yes · no |
How it works
- Install the skills into Hermes, Claude Code or Codex
- At each decision point, the skill asks Jev a typed question
- Agent acts on the answer (route, load skill, open result, keep turn)
- Monitor decisions in the dashboard
Strengths
- Fast, cheap routing and skill selection
- Measured latencies reported
- Honest about a negative handoff result
- No dependencies
Caveats
- Some features lack dedicated benchmarks
- Requires a TypeSafe key
Getting started
git clone https://github.com/kerpopule/hermes-jev-skills ~/hermes-jev-skills && python3 ~/hermes-jev-skills/install.py.At a glance
- Type
- Agent skill
- Status
- beta
- Language
- Python
- Runtime
- agent skills (Hermes, Claude Code, Codex)
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Agent runtime: a Jev-powered skill suite that decides model routing, memory, compaction, skill selection, and computer or browser use for Hermes agents, and also installs under Claude Code and Codex.
yibie/awesome-jevHermes pack, also for Claude Code and Codex: Jev handles model routing, skill choice, search, memory, and compaction. About 0.4 s per routing turn, and about 2.8 s to pick among 377 skills. A measured handoff digest recalled less than the plain transcript, so handoffs keep the dialogue
AnotiaWang/awesome-decision-modelsJev-powered decision toolkit for Hermes, Claude Code, and Codex providing model routing, memory compaction, and dynamic skill selection.
logicrw/awesome-jev-projectsJev-powered model routing, memory, compaction, skill selection, computer and…
wh000wh000/awesome-jev-liveJev-powered model routing, memory, compaction, skill selection, computer and browser use for Hermes agents (also Claude Code and Codex).
hellogumbo/awesome-jevBundle of skills that hand an agent's small decisions to Jev: model routing, skill selection, retrieval filtering, compaction, and computer use, with a routing dashboard. Works with Hermes, Claude Code, and Codex.
MrJev/awesome-jev
Frequently asked questions
How fast is skill selection?
About 2.8 seconds to choose among 377 skills, per the README.
Does Jev write search queries?
No, it picks among candidate queries and results.
Which agents are supported?
Hermes, Claude Code and Codex.
Related projects
One skill and MCP server for Claude Code, Codex and Hermes where Jev gates large-model turns, routes models and guards Git steps.
MCP serverbetaPythonFourteen opt-in Claude Code and Codex skills where Jev makes one bounded choice: which skill, model, test, file or next step.
ProjectbetaPythonClaude Code plugin with three Jev tools: verbatim context compaction, per-edit model routing and per-prompt skill selection.
ProjectbetaPythonClaude Code plugin that routes every subagent spawn through Jev to pick its model, effort and tool profile.
ProjectprototypeOrchestrator that probes which coding agents are installed and logged in, then has Jev pick the best one for a task.
Agent skillbetaPythonHooks for Claude Code, Codex and pi that use Jev to suggest a skill each turn, plus a rule gate for subagent model routing.
HookbetaJavaScript