Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

omp-jevify

Oh My Pi skill for method-first bulk classification: decide when to use Jev, then pilot and freeze a rubric for judge_batch.

Agent skillbeta
View on GitHub ↗

Overview

omp-jevify is an Oh My Pi plugin that ships one skill, jevify. It gives method-first guidance on when semantic judgment is appropriate and, when it is, runs rubric-based bulk classification in the eval kernel with judge_batch backed by Jev. It prefers compilers, search, AST rules or scripts when a deterministic rule decides the answer, and direct reading for small or tightly cross-dependent tasks.

Jev is suggested for repeated, independently judgeable semantic decisions, such as whether changed hunks match a commit message or whether review findings match cited code. The workflow is to define and pilot a rubric, freeze it, cover the declared scope, read flagged sources and resolve unknowns before claiming closure, and calibrate before bulk or destructive action; scores are not treated as proof or permission to edit. It generalises Can Bölük's Jevict idea beyond test pruning. Stock OMP versions inject a conflicting item-count notice that this plugin alone does not change.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this changed hunk explained by the commit message?yes · no
Yes / noDoes this review finding match the cited code?yes · no

How it works

  1. Decide whether a deterministic method or direct reading suffices.
  2. If not, define and pilot a rubric, then freeze it.
  3. Run judge_batch over the full declared scope.
  4. Read flagged items and calibrate before acting.

Strengths

  • Discourages using a model where a rule would do.
  • Rubric piloting and calibration before bulk action.
  • Explicit that scores are not proof.

Caveats

  • Host notice in stock OMP conflicts with its guidance.
  • Requires Jev access for judgments.

Getting started

omp plugin install 'git+https://github.com/ericjuta/omp-jevify.git#v0.3.1', then start a new OMP session and use /skill:jevify.

At a glance

Type
Agent skill
Status
beta
Language
Python
Runtime
Oh My Pi skill plugin
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

When does jevify recommend Jev?

For repeated, independently judgeable semantic decisions with enough context.

Does it always use Jev?

No, it prefers deterministic tools or direct reading when they fit.

How do I invoke it?

Mention jevify in the prompt or type /skill:jevify.

Related projects

  • celesto

    Secure, persistent VM sandboxes for AI agents, with a PR-review example that compares a general model with Jev on candidate findings.

    ProjectproductionPython
  • jev-test-filter

    CLI that scores every test against a git diff with Jev and emits filter arguments for vitest, jest, node, Playwright, cargo, go and rspec.

    ProjectbetaTypeScript
  • Antigravity-mcp-semantic-search-with-TypeSafeAi

    MCP server that finds relevant code with Jev Choice and Noul checks instead of grep sweeps, and audits diffs against task intent.

    MCP serverprototypePython
  • Jev-Code-Reviewer

    Local PR reviewer that has Jev prioritise each change P0-P2 so humans read the important hunks first, with plain-language explanations.

    CLIprototypeTypeScript
  • jev-code

    Gives Claude Code, Codex, Pi and OpenCode five typed Jev tools (classify, check, score, rank, ask) plus a skill on when to use them.

    MCP serverbetaTypeScript
  • jevkit

    Self-hosted Jev server plus CLIs, agent skills and a Pi tool for typed code questions and per-file diff review triage.

    Agent skillbetaPython