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

dsh-jev

DeepSeek Harness plugin adding a jev_evaluate tool for mixed Noul, Choice and Score evaluation of text or JSON.

Projectbeta
View on GitHub ↗

Overview

This dsh-jev plugin adds a jev_evaluate tool to DeepSeek Harness. One request can mix several Noul, Choice and Score questions over text or JSON data and returns the raw answers, probability distributions, confidence and token usage. It calls the HTTP API directly with no TypeSafe SDK or runtime dependencies, and declares the core DSH packages as optional peers so installing it does not pull in a second copy.

It was verified on DSH 0.2.0-rc.2 with Node.js 22. The API key is entered in a plugin details form that uses DSH's official extension point and credentials service; the page shows only whether a key is configured and never echoes it. Credentials resolve from the environment, DSH's credential store, then project or user .env files. Documentation is in Chinese.

What the decision model decides

TypeQuestionAnswers
Yes / noMixed Noul questions over the supplied text or JSON(inferred)yes · no
ChoiceMixed Choice questions(inferred)—
ScoreMixed Score questions(inferred)—

How it works

  1. Install the bundle in DSH and set the key
  2. Agent calls jev_evaluate with state and questions
  3. Plugin posts to Jev's HTTP API
  4. Agent reads answers, distributions and usage

Strengths

  • No SDK or runtime dependencies
  • Key handled by DSH's credential service
  • Mixed question types per call

Caveats

  • Verified only on a specific DSH release candidate
  • Chinese documentation

Getting started

Install the bundle from https://github.com/iptton-ai/dsh-jev/archive/refs/heads/main.tar.gz in DSH's plugin manager, then set the TypeSafe API key in its details page.

At a glance

Type
Project
Status
beta
Language
TypeScript
Runtime
DeepSeek Harness plugin
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What tool does it add?

jev_evaluate, which mixes Noul, Choice and Score questions in one request.

Does it need the TypeSafe SDK?

No, it calls the HTTP API directly.

Which DSH version is supported?

It was verified on DSH 0.2.0-rc.2 with Node 22.

Related projects

  • thinkthen

    Command-line tool and SDK that answers typed questions about text with true/false, a label or a number, for scripts and evals.

    ProjectbetaRust
  • decisions-judge-mcp★ 5

    MCP server exposing a judge tool for typed noul, choice and score questions over JSON state, backed by Jev.

    MCP serverbetaTypeScript
  • dsh-jev-plugin (jackie-cqz)

    DeepSeek Harness plugin exposing Jev noul, choice and score as agent tools, with configurable guardrails and Web UI result cards.

    PluginbetaTypeScript
  • dsh-plugin-jev

    DeepSeek Harness Web GUI plugin giving the agent a jev_ask tool, usage tracking and an optional command guard with explicit caveats.

    PluginbetaTypeScript
  • Jev DSH Decision Engine

    Decision plugin for agent harnesses where Jev recommends tools, skills or owners and scores output, for DeepSeek Harness and iPolloWork.

    Agent skillbetaTypeScript
  • dsh-jev

    DeepSeek Harness bundle registering a jev_ask tool for typed noul, choice and score questions to Jev.

    ProjectprototypeTypeScript