jev SKILL.md skillA

jev from ismaildasci/claude-referee is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add ismaildasci/claude-referee/jev@git:20261015.71bc158

Pins git:20261015.71bc158 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
ismaildasci/claude-referee
Path
plugins/claude-referee/skills/jev/SKILL.md
Versions recorded
8
Latest version
git:20261015.71bc158, committed Oct 15, 2026
First version recorded
Sep 30, 2026
Size
4,887 bytes
Repository stars
8
License
MIT

Hand small, checkable judgements to TypeSafe Jev through the claude-referee CLI instead of judging them yourself. done checks test or lint output against criteria, decide scores 2-6 options against context files, judge runs one yes/no rule over many items, claims (old name verify) checks claims against a source. Use when a check's output decides "done", or when 20+ items need the same rule.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# claude-referee

The CLI is `node "${CLAUDE_PLUGIN_ROOT}/dist/cli.mjs" <command>`. Every command prints one JSON line: `ok`, `verdict`, a few numbers, a `next_step` when there is one, and, for commands that ask Jev, a `receipt`. `<command> --describe` (or `--help`) prints a command's contract, and `--describe` alone lists the commands. `--dry-run` shows the redacted request without sending it.

## When Jev pays off

- A check decides whether the task is done. Let `done --run` run it, or pipe its whole output to `done` without reading it first.
- A choice between 2-6 options whose facts live in files. `decide` reads the files itself.
- One yes/no rule over many items. In the cost model this pays off from about 20 items that aren't in context yet, or about 50 that are (modelled, not measured).
- A single small judgement about text already in context doesn't pay off. Decide it yourself.

## Calls

```bash
node "${CLAUDE_PLUGIN_ROOT}/dist/cli.mjs" done --run test
```
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add ismaildasci/claude-referee/jev@sha256:3471a75fe5b90732

The line at the top of this page pins the label git:20261015.71bc158, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_h75745kmn3qmlp2y.svg)](https://markdownregistry.com/a/art_h75745kmn3qmlp2y)

3 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 8 versions of this file, between Sep 30, 2026 and Oct 15, 2026, about one every 2 days.

versioncommittedcommitsizeaudit
git:20261015.71bc158 latest2026-10-15 71bc158 4,887 BA view · diff
git:20261007.42c38c62026-10-07 42c38c6 4,141 BA view · diff
git:20261007.ea343062026-10-07 ea34306 3,745 BA view · diff
git:20261007.e0dc53b2026-10-07 e0dc53b 2,816 BA view · diff
git:20261005.2e65abc2026-10-05 2e65abc 2,788 BA view · diff
git:20261002.2af0ad02026-10-02 2af0ad0 2,601 BA view · diff
git:20261001.661979d2026-10-01 661979d 2,583 BA view · diff
git:20260930.9865f442026-09-30 9865f44 2,435 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (4887 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

ismaildasci/claude-referee · 8 stars · license MIT · pushed 2026-10-21 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_h75745kmn3qmlp2y
GET https://markdownregistry.com/api/v1/resolve?ref=ismaildasci/claude-referee/jev
GET https://markdownregistry.com/api/v1/blob/3471a75fe5b907327c2e03db5cef10e3b3679cb8d3dc30e385c6c704849480a2

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from ismaildasci/claude-referee

verify SKILL.md skill
ismaildasci/claude-referee · docs/recipes/verify/SKILL.md · Check that the work is really done before committing. Runs the project's test command and pipes its output to…
git:20261007.ea34306 · audit A · 8 stars

Every file in ismaildasci/claude-referee

Other files named jev

jev SKILL.md skill
joetawil7/first-pass · skills/jev/SKILL.md · Sets up, checks or changes the optional Jev judge, the TypeSafe decision model that ship-check asks about review…
git:20260929.5291e17 · audit A · 107 stars
jev SKILL.md skill
francoischastel/jev-code · skills/jev/SKILL.md · Use Jev, TypeSafe's System One classifier, whenever a coding task needs a classifier under the hood: labelling or…
v0.5.0 · audit A · 51 stars
jev SKILL.md skill
pymodel/jev-judge-mcp · src/jev_judge_mcp/skills/jev/SKILL.md · Use when code needs a judgment about natural language that rules or regex cannot make (classify, route, triage…
git:20260927.164b44f · audit A · 48 stars
jev SKILL.md skill
flc1125/skills · skills/jev/SKILL.md · Solve user tasks by dynamically designing and executing workflows that combine code with Jev semantic judgments. Use…
git:20260922.d822388 · audit A · 37 stars
jev SKILL.md skill
calebnewtonusc/chewbacca · skills/jev/SKILL.md · Use Chewbacca's Jev integration for cheap typed decisions, semantic skill selection, classification, scoring, and…
git:20260923.d4201c4 · audit A · 9 stars
jev SKILL.md skill
rashedint32/jev-mcp · skills/jev/SKILL.md · Get typed, calibrated judgments from the connected `jev` MCP tools when the same semantic decision repeats across many…
git:20260920.ad00bf8 · audit A · 7 stars
jev SKILL.md skill
jzkk720/cubecloud-skills-bundle-kit · upstream/jev/SKILL.md · Set up Jev via OpenRouter, native TypeSafe or explicitly chosen simulation; build custom decision integrations and…
git:20260921.ae0b3e2 · audit A · 4 stars
jev SKILL.md skill
lu1sdv/skillsmd · jev/SKILL.md · Use when writing or improving Jev programs — the TypeSafe judgment model behind POST /v1/systemone. Covers primitive…
git:20260922.b382a88 · audit A · 2 stars

Browse by kind, by grade A, or by owner.