Where does deterministic code stop and Jev start?

Python owns IDs, stock math, records, and compatibility, while Jev only resolves ambiguous language such as whether a trip is electrical or a bearing effect.

Can Jev directly update inventory or work orders?

No. The digest states no Jev decision writes to the CSVs; commands return structured proposed actions validated against real CMMS entities.

How are the six decisions evaluated?

A 30-case set per decision with easy, medium, and hard tiers, with captured output in artifacts/results.txt and up to five failures shown per evaluation.

Which decisions still fail and what happens then?

Spare relevance is weakest on hard near-matches and replenishment weakest on medium borderlines, which is why outputs stay as proposals under CMMS controls.

How accurate are the six CMMS-JEV decisions?

On the 30-case evaluations the captured live run reported classification 93.3%, routing 100%, duplicate detection 83.3%, inspection disposition 96.7%, spare relevance 66.7% and replenishment attention 83.3%. Spare relevance is weak on hard near-match cases.

What happens when a CMMS-JEV decision is wrong?

It stays a proposal. Every Jev answer is validated in Python and turned into a non-mutating workflow proposal, and no Jev decision writes to the CMMS CSVs, which is why weak cases cannot change state.

How do I run a CMMS-JEV decision without calling Jev?

Append --dry-run to any individual-decision command to inspect its deterministic context and criteria without calling Jev, and validate the underlying dataset separately with python audit_dataset.py.

CommunityJev Tools & CLIs

CMMS-JEV

Terminal-first synthetic CMMS where Jev makes six bounded semantic calls and Python validates results into workflow proposals.

See highlights

Open github.com

Stars: 0
Language: Python
Published by: Community
Author: Om22210564
Last verified: Oct 2026

Om22210564/CMMS-JEV

User repository on GitHub

View on GitHub

A sample CMMS integration with TypeSafe AI Jev, demonstrating LLM-powered semantic decisions across work orders, routing, duplicates, inspections, spare parts, and inventory using synthetic CMMS data.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

A small, terminal-first synthetic CMMS decision system powered by TypeSafe AI Jev. The CSV dataset is the operational source of truth; Jev supplies constrained semantic judgments where maintenance language is ambiguous. Maintenance reports mix symptoms, equipment context, and operator language. For example, a pump trip can sound electrical while the underlying bearing failure is mechanical. This project separates that semantic judgment from the deterministic CMMS controls that validate and act on it.

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Classification

Used inJev for Software & DevTools

Languages: Python

Highlights

Constrains Jev to classification, routing, duplicate, disposition, spare, and replenishment vocabularies.
Validates teams, entities, thresholds, and quantities in Python before proposing actions.
Demonstrates CP-400 vibration through classification, routing, relevance, and attention steps.
Supports dry-run context inspection, per-case CLI commands, and dataset auditing.

Watch out

Synthetic terminal demo with known weak spots; outputs are proposals inside deterministic controls, never direct writes. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev-deep-dive

Long-form Jev writeup that maps each conclusion to official docs, vendor evals, integrations, community reproductions, and press, separating contract facts from unverified claims.