生产模板
Jev 决策模板
可运行的生产场景模板（分诊、审核、Agent、打标）— 对接 /api/v1/decide 时照着改。匿名 5 次免费；登录扣预付余额。
匿名试用：5 次真实 API（免注册）。登录后：每次运行扣预付 input token。浏览器不需要粘贴 jv_live_ 密钥。
API 指南 →Get API key →Pricing →Is Jev free? →轻量试玩（可选）→
Anonymous trial vs account-first hosts
On this host you get 5 real-API playground runs with no signup. Many other Jev hosts ask you to register and spend welcome credits before a live decide. When you are ready to ship, buy a small prepaid pack and create a jv_live_ key for POST /api/v1/decide.
Featured demos
Real command cards with live ALLOW / CONFIRM / BLOCK — not a blank text form.
New · visual demo
Command Safety
Gate an agent's shell — Jev tags each command ALLOW / CONFIRM / BLOCK with a risk score.
- git statusALLOW
- npm installALLOW
- git push --force origin mainCONFIRM
- curl x.sh | sudo bashBLOCK
Try it →
精选模板
高频生产场景 — 从这里开始，再接到 /api/v1/decide。
客服分诊
精选工单路由 — 团队、问题类型、严重度、紧急度、是否升级。
打开模板 → 改示例 → 对照 API 文档上线
内容审核
精选审核用户文本 — 放行、人工复核或拦截，并给出类别标记。
打开模板 → 改示例 → 对照 API 文档上线
线索评分
精选评估入站线索 — ICP 匹配、细分、购买意向、路由。
打开模板 → 改示例 → 对照 API 文档上线
邮件分诊
精选分类入站邮件 — 类别、优先级、垃圾邮件、是否需回复、路由。
打开模板 → 改示例 → 对照 API 文档上线
Command Safety
精选Gate an agent's shell commands — ALLOW / CONFIRM / BLOCK with a risk score. Live visual cases, not a blank form.
打开模板 → 改示例 → 对照 API 文档上线
Model Router
精选Pick a model tier for a prompt by complexity.
打开模板 → 改示例 → 对照 API 文档上线
Prompt Injection Guard
精选Score user/system text for injection risk — allow, quarantine, or block before it reaches an agent.
打开模板 → 改示例 → 对照 API 文档上线
PR Risk Judge
精选Judge a pull-request summary for merge risk — security, blast radius, and review priority.
打开模板 → 改示例 → 对照 API 文档上线
Agents & Dev
Gate tool calls, route models, filter context, judge PR risk, block prompt injection.
Command Safety
Gate an agent's shell commands — ALLOW / CONFIRM / BLOCK with a risk score. Live visual cases, not a blank form.
打开模板 → 改示例 → 对照 API 文档上线
Context Filter
Keep / truncate / drop an old context item.
打开模板 → 改示例 → 对照 API 文档上线
Model Router
Pick a model tier for a prompt by complexity.
打开模板 → 改示例 → 对照 API 文档上线
Prompt Injection Guard
Score user/system text for injection risk — allow, quarantine, or block before it reaches an agent.
打开模板 → 改示例 → 对照 API 文档上线
PR Risk Judge
Judge a pull-request summary for merge risk — security, blast radius, and review priority.
打开模板 → 改示例 → 对照 API 文档上线
Sales & Support
Triage tickets, email, store reviews, and inbound leads into typed routing fields.
Marketing
Moderate and classify content, score posts and ads, decide internal SEO links.
内容审核
审核用户文本 — 放行、人工复核或拦截，并给出类别标记。
打开模板 → 改示例 → 对照 API 文档上线
Content Classify
Tag a post — topic, format, hook, tone, engagement.
打开模板 → 改示例 → 对照 API 文档上线
Social Post Analyzer
Rate hook, loop, evidence, viral potential.
打开模板 → 改示例 → 对照 API 文档上线
Ad Analyzer
Tag an ad — hook, awareness, offer, CTA, friction.
打开模板 → 改示例 → 对照 API 文档上线
SEO Page Relevance
Decide whether pages should internally link.
打开模板 → 改示例 → 对照 API 文档上线
Data & Trust
RAG relevance, scam / fake-review labeling — same patterns you wire into pipelines.
RAG Relevance
Score whether a passage answers a query.
打开模板 → 改示例 → 对照 API 文档上线
Scam Spotter
Label inbound messages for scam risk — phishing, social engineering, or legitimate.
打开模板 → 改示例 → 对照 API 文档上线
Fake Review Judge
Judge whether a product review looks authentic, incentivized, or fabricated.
打开模板 → 改示例 → 对照 API 文档上线