技能 typesafe-ai

📦

typesafe-ai

Name: typesafe-ai
Author: typesafe-ai

内容修订 r1 安全 ⚙️ 外部命令🌐 网络访问

Build Typed AI Decisions with TypeSafe

Many application features need semantic judgments, but free-form model responses are difficult to combine reliably in code. TypeSafe helps developers design small typed judgments and compose them into routing, ranking, extraction, verification, and interactive workflows.

支持: Claude Codex Code(CC)

🥉 78 青铜

使用我的智能体安装

将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。

智能体请求

Review the Skillstore skill "typesafe-ai" from https://skillstore.io/skills/typesafe-ai-typesafe-ai.md and its manifest at https://skillstore.io/api/skills/typesafe-ai-typesafe-ai/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

你的智能体仍应展示其计划，并按安全策略请求必要确认。

Agent 可读资源

当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。

Markdown 详情 GET /skills/typesafe-ai-typesafe-ai.md 已签名 manifest GET /api/skills/typesafe-ai-typesafe-ai/manifest 已签名 lockfile GET /api/skills/typesafe-ai-typesafe-ai/lockfile

测试它

正在使用“typesafe-ai”。 A support application must route incoming messages to a known team and fill the team-specific fields.

预期结果:

Use a routing judgment to select a team and a separate typed judgment for branch-specific arguments. Ask only the questions relevant to the selected branch, then let application code execute the handoff.

正在使用“typesafe-ai”。 A search workflow has many retrieved candidates and needs reliable context for an answer.

预期结果:

Retrieve candidates in code, judge relevance with comparable scores, select useful evidence, and send uncertain or failing cases to verification or human review.

正在使用“typesafe-ai”。 A product team wants to replace a broad prompt that returns parsed text.

预期结果:

Split the broad request into narrow judgments with explicit criteria, named state fields, typed outputs, a no-match result, and tests for representative cases.

安全审计

安全

v1 • 2026/9/17 打开版本化报告

All 26 static findings are false positives. The flagged backticks are Markdown examples, the URLs are documentation links, and the fallback guidance does not perform system reconnaissance; no semantic security findings were identified.

已扫描文件

172

分析行数

复核项

已忽略误报

风险因素

⚙️ 外部命令 (3)

SKILL.md:34 SKILL.md:37 SKILL.md:110

🌐 网络访问 (22)

最新一次完成的静态与语义审计未检测到已确认的安全发现。这并不能证明该 Skill 没有副作用。

审计者: claude

分享与引用此报告

分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。

打开版本化报告

复制报告链接

https://skillstore.io/skills/typesafe-ai-typesafe-ai/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/typesafe-ai-typesafe-ai/security.svg)](https://skillstore.io/skills/typesafe-ai-typesafe-ai?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/typesafe-ai-typesafe-ai?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/typesafe-ai-typesafe-ai/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/typesafe-ai-typesafe-ai.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

学术引用 (APA · BibTeX · CFF)

APA 引用

typesafe-ai. (2026). typesafe-ai security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/typesafe-ai-typesafe-ai/audits/1

BibTeX 引用

@techreport{typesafe-ai-typesafe-ai-typesafe-ai-2026,
  author = {typesafe-ai},
  title = {typesafe-ai security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/typesafe-ai-typesafe-ai/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "typesafe-ai security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "typesafe-ai"
date-released: "2026-09-17"
url: "https://skillstore.io/skills/typesafe-ai-typesafe-ai/audits/1"
identifiers:
  - type: other
    value: "skillstore:typesafe-ai-typesafe-ai:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore 评分

评分依据证据置信度: 中

架构

可维护性

内容

社区

规范符合性

你能构建什么

Plan semantic application features

Turn a product behavior into focused judgments, typed outputs, and bounded application actions.

Design evidence and review workflows

Combine retrieval, relevance judgments, confidence thresholds, verification, and human escalation.

Improve an existing model integration

Replace broad prompt-and-parse steps with reusable primitives, explicit state, and testable composition.

试试这些提示

Find the right judgment

I am building [feature]. It must [observable behavior]. Identify the smallest semantic judgment it needs, the relevant state, and a suitable TypeSafe primitive. Keep deterministic rules in code.

Choose a primitive

For this decision, compare Choice, Noul, and Score: [decision]. Define what each output means, when each primitive fits, and what no-match or uncertainty outcome the application needs.

Compose a workflow

Design a TypeSafe workflow for [goal]. Separate observed state, inferred state, policies, and execution. Identify independent questions that can run together, dependent follow-up steps, thresholds, and human escalation.

Audit a production design

Review this proposed semantic workflow: [design]. Check question scope, candidate coverage, state freshness, confidence interpretation, policy handling, latency and cost, failure categories, test cases, and the boundary between model judgments and application code.

最佳实践

Start from the application behavior, then define the smallest judgment that supports it.
Keep rules, calculations, lookups, execution, and authorization in deterministic code.
Test judgments and resulting behavior with representative data, thresholds, and failure cases.

避免

Do not use a model judgment for exact rules, arithmetic, lookups, or permission enforcement.
Do not treat confidence as proof of correctness or reuse example thresholds without validation.
Do not send stale inferred state into actions without checking the current application state.

常见问题

What is TypeSafe used for?

It helps developers add focused semantic judgments to software for routing, ranking, extraction, verification, and interactive decisions.

Which primitive should I choose?

Use Choice for one option, Noul for whether a condition holds, and Score for a degree along a defined dimension.

Does this skill write integration code?

It provides design guidance and points to API and SDK documentation. It does not implement or execute an integration.

How should I handle uncertainty?

Define thresholds using target data and consequences, include no-match outcomes, and escalate uncertain or failing cases when appropriate.

Can TypeSafe replace application logic?

No. Keep deterministic rules, calculations, lookups, execution, authorization, and workflow control in code.

What should I test?

Test representative inputs, state freshness, questions, candidates, outputs, composition, thresholds, service failures, and observed application behavior.