skills.rest
PedroAlvarado
PedroAlvaradoCOMMUNITY
Explore all skills by this author →
GitHub

jev-scout

Inspect a repository and rank where TypeSafe Jev decision models create value.

Updated Sep 23, 2026
Search Tags:#repository analysis#opportunity discovery#decision models#llm cost reduction#shadow mode#jev#typed judgments
One-click install
npx skills add https://github.com/PedroAlvarado/jev-scout --skill jev-scout-pedroalvarado
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: jev-scout
Source: https://github.com/PedroAlvarado/jev-scout/tree/main/skills/jev-scout
Command: npx skills add https://github.com/PedroAlvarado/jev-scout --skill jev-scout-pedroalvarado

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Teams adopting TypeSafe Jev decision models struggle to identify which judgments in an existing codebase are worth converting, and often miss opportunities beyond obvious LLM-call replacements. This Skill reads a repository and produces an evidence-grounded, ranked map of substitution, augmentation, and new-capability opportunities. ## Core Features & Use Cases - Decision inventory scanning: A bundled TypeScript scanner lists model calls, label-seeking prompts, keyword heuristics, first-match picks, fixed cutoffs, and manual review steps with file and line references. - Failure-record analysis: Git history, heuristic churn, tests, and TODOs reveal which decisions keep breaking and where labeled evaluation data already exists. - Ranked opportunity report: Candidates are challenged against a fit test, scored on business leverage and risk, and the top three receive Jev Choice/Noul/Score decision contracts plus shadow-mode experiment designs. - Use Case: Point your coding agent at a SaaS repository and ask where Jev would pay off; receive a value map, a decision inventory, and a ranked opportunity report with integration points and shadow experiments, without any code being changed. ## Quick Start Open your agent at the repository root and ask it to find where Jev decision models would create the most value in this codebase.

Frequently Asked Questions about jev-scout

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find Jev use cases in my codebase?▼

Run the skill from the repository root and ask where Jev would pay off. It maps the value chain, scans for model calls and heuristics, reads git fix history, and delivers a ranked opportunity report with decision contracts for the top picks.

What kinds of opportunities does Jev Scout detect?▼

It detects three families: substitution of existing model calls or keyword rules, augmentation of implicit decisions like first-match picks and fixed top-k cutoffs, and new capabilities that become practical when typed judgments are fast and cheap.

Does Jev Scout require a Jev API key or modify my code?▼

No API key is needed and the skill is analysis-only. It changes no code and writes nothing outside its report; the only optional network reads are TypeSafe's public documentation pages.

What runtimes does the bundled scanner support?▼

The scanner runs on Bun or Node.js 22.6+ with built-in type stripping, or an already-installed local tsx. Without any of these, the skill falls back to rg, git grep, and direct file reads.

Which coding agents are compatible with this skill?▼

It works in any Agent Skills host with shell and file access, including Claude Code, Codex, and Cursor. It avoids provider-specific commands and resolves bundled files relative to its own SKILL.md.

When is Jev not a good fit for a decision?▼

Jev is not recommended for exact arithmetic, deterministic lookups, schema validation, permissions, open-ended generation, multi-step reasoning, or rare decisions where integration overhead dominates. Dropped candidates are recorded in the report's Not a Jev fit section.

Related Skills

Explore More Software Engineering Skills→
parisgroup-aiparisgroup-ai
1

sync-skills

Sync shared skill directories from a remote Git repository into local Claude Code and Codex skill folders.

Official
Intermediate
kai-koukai-kou

typesafe-ai

Compose typed AI judgments and probabilities into application code using TypeSafe System One models.

Community
Intermediate
Jin9Jin9

performance-cost-review

Define performance and cost budgets with baselines, headroom, and ordered optimization levers.

Community
Intermediate
4lerman4lerman

llm-cost-optimizer

Analyze AI API spending and implement model routing and prompt caching strategies.

Community
Advanced