Token导航 LogoToken导航TokenDH.com
研究检索external-servicegithub未标认证来源可访问许可证需确认审计通过

ascn-operatorascn 运算符

Agent Skill

ascn-operator 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。

总安装

729

周安装

31

GitHub Stars

公开资料未说明

下载量

255
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:ascn-operator(ascn 运算符)
来源仓库:https://github.com/ascnai/skills
仓库路径:skills/ascn-operator
安装命令:
npx skills add https://github.com/ascnai/skills --skill ascn-operator
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/ascnai/skills --skill ascn-operator

简介

ascn-operator 用于操作 ASCN 工作流,包括创建、修复、导出、删除与发布插件等生命周期管理。

  • 它通过 MCP control.* 工具实现确定性操作,支持审计与诊断,提升运维效率。
  • 适用于工作流变更与插件发布,需指定意图与必要输入参数。
  • 安装来自 GitHub,注意 workspace ID 与权限范围,避免误操作关键资源。
  • 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。

SKILL.md

ASCN Operator

Operate ASCN workflows using MCP control.* tools with deterministic, auditable outcomes.

When To Use

Use this skill for:

  1. workflow lifecycle changes (create|repair|patch|export|delete)
  2. plugin publication (publish_plugin)
  3. operator diagnostics (explain)

Do not use this skill to design brand-new handlers/triggers. Delegate that to ascn-integrations.

Required Inputs

The operator MUST collect:

  1. intent (create|repair|patch|export|publish_plugin|delete|explain)

Intent-specific required inputs:

  1. workflow_id for patch|repair|export|delete
  2. plugin_name and handlers[] for publish_plugin
  3. publish_plugin handlers MUST be user-defined and match User.<Handler>

Recommended inputs:

  1. success criteria (expected tool name, run status, or plugin visibility)
  2. blueprint_preference (linear|fanout|conditional|retryable_http|tool_export)
  3. constraints (latency_slo_ms, throughput_rps, idempotency_requirement)

If required inputs are missing, stop and return input_validation_error.

Dependency And Discovery Policy

At task start, the operator MUST run:

  1. control.docs.get
  2. tool discovery check for required surface:

- control.registry.list - control.registry.details - control.registry.resolve_options - control.workflows.list - control.workflows.describe - control.workflows.validate - control.workflows.create - control.workflows.patch - control.workflows.activate - control.workflows.delete - control.tools.list_exports - control.tools.ensure_export - control.runs.list - control.runs.details - control.plugins.create_plugin - control.plugins.validate_definition - control.plugins.update_plugin - control.plugins.get - control.plugins.list

Connectivity defaults (use as hints, not hard requirements):

  1. transport: streamable_http
  2. dependency id: workspace-mcp-gateway
  3. URL hint: https://nocode.ascn.ai/mcp
  4. secret hint: mcp_gateway_token
  5. auth header: Authorization: Bearer <token> (must match secret value)
  6. token generation URL: https://ascn.ai/no-code/mcp-list

If discovery fails, do not mutate. Return dependency_failure with missing tools and connection hints.

Capability Gap Policy

After dependency checks, classify capability in this order:

  1. inspect workflows/exports (control.workflows.list, control.workflows.describe, control.tools.list_exports)
  2. inspect handlers/triggers (control.registry.list, control.registry.details)
  3. classify as one of:

- sufficient - missing_handler - missing_trigger - missing_auth_capability - schema_or_contract_gap

If not sufficient, the operator MUST NOT invent handler names. It MUST pause mutations and delegate to skills/ascn-integrations/SKILL.md.

Blueprint Selection

For create|patch|repair|export, choose exactly one:

  1. linear
  2. fanout
  3. conditional
  4. retryable_http
  5. tool_export

Then enforce topology consistency with the selected blueprint.

Execution Flow

Common authoring pipeline:

  1. select blueprint
  2. schema-lock from control.registry.details (required fields, defaults, secret-backed fields)
  3. build minimal valid draft
  4. run reference safety gate:

- every $node[...] has a reachable path - no raw $... directive without ={{...}} - no secret literals

  1. validate (control.workflows.validate)

- payload MUST be wrapped as {"workflow": {...}}

  1. mutate (create or patch)
  2. activate (control.workflows.activate)
  3. expand only when required by scope

Intent call order:

create

  1. control.workflows.list
  2. control.registry.list
  3. control.registry.details
  4. authoring pipeline with create

patch|repair

  1. control.workflows.list
  2. control.workflows.describe
  3. control.registry.details
  4. authoring pipeline with patch

export

  1. control.workflows.describe
  2. control.tools.list_exports with expose_mcp_only=false
  3. control.tools.ensure_export (must include output_path)
  4. control.workflows.validate

- payload MUST be wrapped as {"workflow": {...}}

  1. control.workflows.activate
  2. smoke invoke exported tool with minimal payload
  3. control.runs.list
  4. control.runs.details on failure/unexpected status

publish_plugin

  1. validate every handler matches User.<Handler>; reject system handlers (for example Telegram.*)
  2. control.registry.details for each handler
  3. collect existing plugin examples using control.plugins.list with include_definition=true (never query DB directly)
  4. use control.plugins.get for deterministic single-definition reads when a definition_id already exists
  5. for fields with options_source, resolve via control.registry.resolve_options
  6. when editing params_ui, apply conditional patterns and validation checklist from skills/ascn-integrations/SKILL.md (Conditional UI Recipes, Conditional UI Validation Checklist)
  7. preflight with control.plugins.validate_definition
  8. control.plugins.create_plugin
  9. optional control.plugins.update_plugin
  10. control.plugins.list verify plugin is visible

delete

  1. control.workflows.describe
  2. summarize destructive impact
  3. control.workflows.delete with confirm=true

Authoring Standards

  1. Activity IDs MUST be unique.
  2. Every edges[].to MUST reference an existing activity.
  3. Trigger entry edges SHOULD be explicit for clear starts.
  4. $json MUST be used only for current node input.
  5. Upstream reads MUST use $node['id'].json.field with graph reachability.
  6. Dynamic expressions and secrets MUST use ={{...}}.
  7. Credentials MUST NOT be hardcoded.

Retry And Stop Rules

  1. Operation key format: {intent}:{workflow_id|workflow_name}:{payload_hash}
  2. Retry only transient failures (timeout, 5xx, transport unavailable), max 3 attempts with exponential backoff.
  3. Never auto-retry validation, schema, or export-conflict errors.
  4. Never delete without explicit confirm=true.
  5. Never mutate when dependency checks fail.

Output Contract

Every completion MUST return:

{
  "intent": "create|repair|patch|export|publish_plugin|delete|explain",
  "workflow_id": "uuid-or-null",
  "selected_blueprint": "linear|fanout|conditional|retryable_http|tool_export|null",
  "capability_status": "sufficient|missing_handler|missing_trigger|missing_auth_capability|schema_or_contract_gap",
  "actions": [
    {
      "step": "string",
      "tool": "control.*",
      "status": "completed|skipped|failed"
    }
  ],
  "validation": {
    "validated": true,
    "activated": true
  },
  "verification": {
    "smoke_tested": true,
    "latest_run_status": "COMPLETED|FAILED|RUNNING|UNKNOWN",
    "run_id": "string-or-null",
    "trace_id": "string-or-null"
  },
  "plugin": {
    "plugin_name": "string-or-null",
    "visible_in_plugins_list": true
  },
  "failure": {
    "class": "none|input_validation_error|dependency_failure|validation_failure|runtime_failure|capability_gap",
    "message": "string"
  },
  "next_action": "string"
}

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

38.07%
按下载量换算97

Claude

29.96%
按下载量换算76

Cursor

19.92%
按下载量换算51

Gemini CLI

9.28%
按下载量换算24

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

external-service

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills