kerpopule/hermes-jev-skills
1,015 stars · Last commit 2026-10-02
Jev-powered model routing, memory, compaction, skill selection, computer and browser use for Hermes agents (also Claude Code and Codex)
README preview
# Hermes Jev Skills **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. That is what [Jev](https://docs.typesafe.ai) is. It is TypeSafe's decision model. **It never writes text.** You give it a state and typed questions (pick one, score this, yes or no) and it answers with a calibrated confidence. This repo wires that into an agent's day.  *`jev dashboard`. The profiles, paths and decisions here are a demo home; the pools are a real working set. `python3 scripts/demo_home.py` builds a home where nothing is real, which is where the next one comes from. One switch for Jev routing, every pool as a tier-by-work-kind grid, and each decision as it happens (tier, work kind, model, which pool it came from, confidence, latency).* ## What Jev decides, and what it costs | Skill | What Jev decides | Measured | |---|---|---| | **Model routing** | Which model is good enough for this turn, from every model you can call | ~0.4 s per turn | | **Search** | Which of the results you fetched are worth opening, whether the evidence answers the question, and which of your candidate queries to run next. Jev never writes a query | ~1.9 s per round (median of five live runs, two requests: rank, then sufficiency and the pick) | | **Social research** | Which discovered posts are worth opening and, after a local evidence-floor check, whether a bounded projection answers the question. Depth and source linkage are enforced locally, not by Jev | Reuses `jev search`; no social-specific benchmark yet | | **Memory** | Which retrieved passages are worth reading, and which contain hidden instructions | 60 passages per request, up to 480 per call; an injection screen runs locally even when Jev is down |