jev-prompt-enhancer
Agent skill where Jev scores a task prompt's clarity, the agent refines it up to three times until it reaches 80/100, then builds.
Overview
Jev Prompt Enhancer is an agent skill that gives a coding agent a clearer task before it writes code. Jev evaluates the prompt and pinpoints where the task, context, constraints or expected output need more clarity; the skill uses that feedback to sharpen the instructions while preserving the user's intent and scope, then continues with the requested work.
The loop is measure, refine, recheck, build. It stops when the prompt reaches 80/100 by default (or a chosen threshold) or after three enhancement attempts, in which case it keeps the highest-scoring version. It targets bug fixes, feature work and review requests, aiming for fewer rounds of 'that's not what I meant'. It needs Bash and Node 18+ with no npm dependencies, reads the key from the environment or a strictly parsed .env file, and writes debug logs to the workspace.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How clear and actionable is this prompt (0-100)? | — |
| Yes / no | Is the expected output clearly specified?(inferred) | yes · no |
How it works
- Invoke $jev-prompt-enhancer with a task
- Jev scores the prompt and gives focused feedback
- The agent refines weak spots while keeping intent
- Recheck until 80/100 or three attempts, then proceed
Strengths
- Fewer clarification rounds with the agent
- Bounded loop with a clear stop rule
- No npm dependencies; safe .env parsing
Caveats
- A clarity score is not a guarantee of correct results
- Requires a TypeSafe key
- Adds latency before work starts
Getting started
npx skills add joacotornello/prompt-enhancer, then export JEV_API_KEY and invoke $jev-prompt-enhancer.At a glance
- Type
- Project
- Status
- beta
- Language
- JavaScript
- Runtime
- Agent skill (Bash and Node)
- Has code
- yes
Use cases and integrations
How curated lists describe it
Enhance your prompts using Jev by giving your coding agent a clearer task before it writes the first line.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
A quality score for the prompt and where it needs more clarity.
When does it stop refining?
At 80/100 by default or after three attempts, keeping the best version.
Will it change what I asked for?
It is designed to preserve your intent and scope while improving clarity.
Related projects
Resume review app and MCP server where a Cloudflare Worker asks Jev typed questions and composes a JevScore, generally or per job.
AppbetaTypeScriptAgent skill where Jev flags 35 AI-writing patterns per paragraph and the model edits only the flagged spots.
ProjectbetaPythonChinese controlled-writing agent skill for technical reports with nine rules, using static checks plus a Jev semantic gate for certainty.
Agent skillbetaRussian-language Letta agent skill that uses Jev as a fast external judge, with rules for treating mid-range probabilities as 'unsure'.
ProjectprototypePythonMCP server over Apple developer docs and WWDC transcripts that uses hybrid search, then Jev relevance scores to rank results.
MCP serverproductionTypeScriptLocal long-term memory for coding agents with an optional Jev reranker that orders recalled memories.
ProjectproductionTypeScript