What is the north-star skill?

A Hermes Agent skill that interviews the owner until a written finish line exists, refuses a result that cannot be judged, and hands back the star, a generated run prompt and a gate that decides whether the work is finished.

Why do long autonomous runs need it?

The README says they fail in one of two ways: they stop early and call it done, or they never stop. Both come from the same missing thing, a written finish line with states someone can point at.

How does Jev judge requirements?

Requirements no script can decide — like the round-up arrived as a whole share — go to Jev, which answers with a probability and a calibrated confidence instead of prose. The gate stays code and never reads a model's opinion of its own progress.

Do I need an API key or dependencies?

The deterministic half is pure Python, stdlib only, with no key needed. Jev is wired in as the default judge when TYPESAFE_API_KEY is present, and the README says the gate states plainly when it is not.

How do I install north-star?

hermes skills install with the SKILL.md URL; hermes skills inspect previews it first. That copies the skill document, references/ and templates/ but not scripts/, so clone the repo once to have the tool on disk.

Can I use it without Hermes?

Yes. Any agent that can run a shell command can use it: the skill is one markdown file plus two stdlib Python scripts, installed by pointing the agent at the SKILL.md file.

What does a star contain?

JSON with a goal, for_whom, requirements, per-requirement checks, out_of_scope items and stop rules. The checks turn each requirement into a command or a Jev question, and exit 1 names the weakest requirement still unproven.

Jev directory

CommunityJev Agents & MCP

North Star

A Hermes Agent skill that turns an intention into a written finish line: machine-checkable requirements run as shell commands, fuzzy ones go to Jev for calibrated probabilities, and exit 1 names the weakest unmet requirement.

See highlights Open github.com

Stars: 2
Language: Python
License: MIT
Reading time: 2 min
Published by: Community
Author: poponline63
Last verified: Oct 2026

poponline63/hermes-jev-north-star

User repository on GitHub

View on GitHub

Hermes Agent skill whose north-star gate is judged by Jev (TypeSafe System One): turn an intention into a checkable finish line, generate the run prompt, and let Jev rank what is still unproven.

Forks: 1
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity catalog-import logicrw hermes agent-skills definition-of-done

Works forJev for Verification Jev for Decision gating

Used inJev for Software & DevTools

Languages: Python
Platforms: CLI
Capabilities: Open sourceRuns locally

Highlights

The gate is two-layered: machine-checkable requirements run as plain shell commands, and fuzzy ones go to Jev for a probability and calibrated confidence.
Exit 0 means every requirement is met; exit 1 keeps going and names the weakest requirement, so the gate stays code.
Hands back three artifacts: the star, the generated run prompt, and the gate that decides whether the work is actually finished.
Pure Python, stdlib only, no dependencies; everything except Jev runs offline, and the gate says plainly when TYPESAFE_API_KEY is absent.
Install from a SKILL.md URL with hermes skills install; it copies the skill document, references/ and templates/, but scripts/ must come from a clone.

Quickstart

bash

hermes skills install https://raw.githubusercontent.com/poponline63/north-star/main/SKILL.md
git clone https://github.com/poponline63/north-star ~/north-star
python3 ~/north-star/scripts/north_star.py --help

Watch out

Copy reviewed against the truncated repo README on 2026-10-15; the JSON example cuts off in stop rules. The README states the project is not an official Nous Research or TypeSafe project, and the offline and judged-by-Jev claims are the README's own.

Questions about this entry

Short answers drawn from the linked project or page.

From the Jev guides

Original write-ups on what this entry is for.

MCP tool routing: Jev in front of the tool call

Oct 16, 2026

Tool menus can outgrow the context that reads them. The catalog's Jev-based routers and gates narrow the list: benchmarked tool search, per-tool scoring, fail-closed approvals, and WebMCP in the browser.

Jev Agents & MCP#mcp#tool-routing#agents

Read article

Approval policies for coding agents: hooks, guards, and gateways