typed Jev question
An agent skill for **Claude Code** and **Codex**. Point it at a
repository and it surfaces the semantic decisions — routing, triage, relevance,
escalation — that could become a **Choice**, **Score**, or
**Noul**. Ranked, source-grounded, read-only.
Three primitives it maps to #
A search lead is only jevable when the judgment is semantic, bounded, and worth measuring.
Choice
Select one supplied option.
dispatch · intent · categorization
Score
Locate content on ordered, described levels.
severity · quality · relevance
Noul
Evaluate one defined yes/no proposition.
flags · matching · escalation
How the audit works #
A seven-step read-only pipeline. Nothing runs, nothing is sent to Jev, no paid inference to discover candidates.
One skill, both platforms #
Claude Code and Codex read the same SKILL.md. Pick your install.
Claude Code copy into skills
git clone https://github.com/AltSlate-Labs/jevable-code.git
cp -r jevable-code/skills/find-jevable-code \
~/.claude/skills/
Codex install the plugin
codex skill install \
https://github.com/AltSlate-Labs/jevable-code.git
Then, in a session with the skill available:
Use find-jevable-code to audit this repository for decisions suited to
Jev Choice, Score, or Noul, and rank the opportunities with code evidence
and concrete question contracts.
See the worked example — a toy keyword router and the audit report the skill produces for it.
source & further reading
altslate-labs.github.io — original article