Token导航 LogoToken导航

Agent Skill · 研究检索

coral-create-source-spec(珊瑚创建源规范)

coral-create-source-spec 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

coral-create-source-spec 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 66,424。

总安装 264周安装 11GitHub Stars 公开资料未说明下载 88核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:coral-create-source-spec(珊瑚创建源规范)
来源仓库:https://github.com/withcoral/skills
仓库路径:skills/coral-create-source-spec
安装命令:
npx skills add https://github.com/withcoral/skills --skill coral-create-source-spec
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/withcoral/skills --skill coral-create-source-spec

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于创建或修复 Coral 数据源的 YAML 规范文件,确保可被查询和验证。适用于外部开发者独立编写数据源配置,支持本地 lint、添加与测试流程。默认生成可独立运行的源定义,需配合 coral source lint 等命令校验格式。

安装方式:GitHub,命令为 npx skills add https://github.com/withcoral/skills --skill coral-create-source-spec。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/coral-create-source-spec

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/withcoral/skills --skill coral-create-source-s…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

coral-create-source-spec 怎么安装?

在终端执行 npx skills add https://github.com/withcoral/skills --skill coral-create-source-spec。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

coral-create-source-spec 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

coral-create-source-spec 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

coral-create-source-spec 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

coral-create-source-spec 用的人多吗?

公开数据是累计安装 264 次,近一周安装 11 次,下载 88 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

coral-create-source-spec 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
coral-create-source-spec

珊瑚创建源规范 · 本页

264—接触敏感数据未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Create Source Spec

Use this skill when the task is to author or repair a Coral source spec.

Goal

Produce a valid, queryable Coral source spec that works with:

  • coral source lint <path>
  • coral source add --file <path>
  • coral source test <name>
  • coral sql
  • coral.tables and coral.columns
  • coral.inputs for source variables and secrets

Default Mode

Default to standalone source authoring for external developers.

That means:

  • create a YAML source spec file
  • lint it early with coral source lint <path>
  • add it to Coral with coral source add --file <path> when you need to exercise it as a source
  • validate by querying it
  • iterate until the shape is correct

Only switch to repo-bundled layout when the user is explicitly editing the Coral repo.

Output Modes

  • External authoring:

- create a standalone source spec such as ./my-source.yaml - validate structure with coral source lint./my-source.yaml - load it with coral source add --file./my-source.yaml when you need to query it through Coral

  • Coral repo contribution:

- write the source spec to sources/<name>/manifest.yaml - add representative test_queries for a basic smoke/connection check of the source - validate with coral source test <name> and repo checks

Workflow

  1. Read the provider API docs or inspect the local dataset.
  2. Start with one small table and a few columns.
  3. Define:

- source metadata - backend - base URL or file location - auth - variables and secrets - tables - filters - response extraction - pagination - typed columns

  1. Lint the source spec:

- coral source lint <path>

  1. Validate the source in the right mode:

- standalone specs: coral source add --file <path> and inspect with coral sql - coral source add is non-interactive by default: each input key is read from the matching environment variable. Export required variables and secrets before running, or pass --interactive to be prompted. - named or repo-bundled sources: coral source test <name>

  1. Inspect the exposed shape:

- inspect coral.tables - inspect coral.columns - inspect coral.inputs to verify variables, secrets, defaults, hints, and required flags

  1. Query representative tables with coral sql.
  2. If you are relying on coral source test, make sure test_queries gives you a basic smoke/connection check for the source.
  3. Refine the spec and repeat.

Authoring Rules

  • Start small and expand table coverage incrementally.
  • Use the source manifest schema as both inspiration for authoring and validation of structure: https://github.com/withcoral/coral/blob/main/crates/coral-spec/src/schema/source_manifest.schema.json
  • Use source variables for non-secret configuration.
  • Use source secrets for credentials.
  • Keep table names stable and SQL-friendly.
  • Mark filters as required only when the API truly requires them.
  • Prefer explicit pagination when the API shape is known.
  • Verify pagination with actual row fetches, not only COUNT(*).
  • Add or update test_queries when you want coral source test to perform a basic smoke/connection check.

Metadata UX Rules

Use these rules for top-level source metadata so source discovery and setup are consistent.

description

  • Start with Query....
  • Make the first sentence capability-first: list the key entities users can query.
  • Preferred template:

- Query <entities> from <Provider> (<Cloud or self-hosted when relevant>).

  • Keep description focused on data coverage, not setup steps.
  • Do not use vague phrasing such as:

- REST API v3 - OpenAPI provider - ... and more

  • Move auth/setup/permission details to input hints, not description text.

Input hints (inputs.<KEY>.hint)

Each hint should tell the user:

  • what value is expected
  • how to obtain it
  • minimum scope/permission guidance
  • one concrete format/example when useful

Specific guidance:

  • For URL/base inputs:

- say what the default means - include at least one concrete example - include self-hosted guidance when supported

  • For secrets:

- name the exact credential type (API key, PAT, application key, etc.) - include format constraints when relevant (for example, token prefixes) - include least-privilege scope guidance

  • For derived secrets (for example Basic auth blobs):

- include a short shell example (for example a Base64 command)

  • Prefer stable documentation links.

- Use official docs links and stable settings pages. - Avoid brittle click-path instructions as the primary guidance.

Keep hints concise and directly actionable.

Validation Loop

Use this loop during authoring:

# Export any required inputs first (key matches the input `key` in the spec),
# or pass --interactive to be prompted.
coral source lint ./my-source.yaml
coral source add --file ./my-source.yaml
coral sql "SELECT * FROM coral.tables WHERE schema_name = 'my_source'"
coral sql "SELECT * FROM coral.columns WHERE schema_name = 'my_source'"
coral sql "SELECT key, kind, value, default_value, hint, required, is_set FROM coral.inputs WHERE schema_name = 'my_source' ORDER BY key"

For repo-bundled or already-named sources, add test_queries for a basic smoke/connection check and run:

coral source test my_source

Then run targeted table queries until the source behaves correctly.

HTTP Sources

For HTTP-backed sources:

  • define backend: http
  • define base_url
  • define auth headers
  • define request path, query, and body only where needed
  • define response rows_path
  • define pagination explicitly when the provider pattern is known
  • define typed columns
  • add test_queries once you know which simple query or queries should confirm the source basically works

Read references/http-source-checklist.md when you need table-shape and pagination guidance.

If your HTTP source uses an Authorization header with a prefix (e.g. Authorization: Bearer <token>), you can use a secret input for the token and define the header as a template:

auth:
  headers:
    - name: Authorization
      from: template
      template: Bearer {{input.FOOBAR_API_KEY}}

Local Data Sources

For local file-backed sources:

  • define the file backend
  • define the source location
  • define file selection patterns if applicable
  • define typed columns

Deliverable

Report:

  • source spec path
  • lint / add / test commands used
  • validation commands run
  • assumptions made
  • blocked or unverified endpoints

平台分布

Codex

38.32%
按下载量换算34

Claude

29.2%
按下载量换算26

Cursor

19.32%
按下载量换算17

Gemini CLI

10.31%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills