ask-jev-skill
Hermes skill that calls Jev as a typed tiebreaker, returning Choice, Score or Noul probabilities so the agent can act or escalate.
Overview
askjev is a small skill for the Hermes agent (and other agents) that calls TypeSafe Jev (jev-latest) as a typed tiebreaker. Jev is not a chat model; it returns Choice, Score or Noul answers with probabilities so the agent can either act or escalate.
It installs by copying SKILL.md and scripts into the Hermes skills directory, reads TYPESAFE_API_KEY from the Hermes .env, and can be disabled with ASKJEV_DISABLED=1. A script runs choice questions with named options, and an eval script runs a live reliability suite. SKILL.md explains when to call Jev, confidence gates and pitfalls.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which path? | a · b |
| Yes / no | Bounded yes/no judgment for the agent(inferred) | yes · no |
How it works
- Agent hits a bounded judgment
- Skill calls askjev.py with state, question and options
- Jev returns probabilities
- Agent acts above the confidence gate, otherwise escalates
Strengths
- Simple drop-in skill
- Easy kill switch
- Live reliability suite
Caveats
- Needs a TypeSafe key
- Small scope
Getting started
Copy SKILL.md and scripts into $HERMES_HOME/skills/autonomous-ai-agents/askjev and add TYPESAFE_API_KEY to $HERMES_HOME/.env.At a glance
- Type
- Agent skill
- Status
- beta
- Language
- Python
- Runtime
- Hermes agent skill
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A skill for Hermes and other agents to query TypeSafe Jev for bounded option judgments and confidence escalations.
logicrw/awesome-jev-projectsSkill for Hermes, and other agents, to ask typesafe's jev.
hellogumbo/awesome-jev
Frequently asked questions
What is it for?
Letting an agent ask Jev for a typed tiebreaker and act or escalate based on confidence.
How do I disable it?
Set ASKJEV_DISABLED=1 or delete the skill directory.
Does it work outside Hermes?
The listing says other agents can use it too.
Related projects
Portable decision layer for agent harnesses: Jev picks from host-supplied capabilities while the host keeps permissions and execution.
ProjectbetaTypeScriptDeepSeek Harness plugin that registers a native jev_decide tool and declares its own policy for when the agent should ask Jev.
PluginbetaJavaScriptRust MCP server for Codex that hands classification, relevance, candidate selection and evaluation to Jev with accept/verify policies.
MCP serverbetaRustJev decision layer for agents: seven MCP tools, an embeddable library and escalate-only Claude Code hooks at harness boundaries.
MCP serverbetaTypeScriptLocal MCP server that runs versioned packs of typed Jev judgments, with a closed tool catalogue and a key configured once.
MCP serverbetaPython router that has Jev classify Grok Bot requests before costly work, returning actions like reuse_cache, stop_retry or ask_human.
Agent skillprototypePython