Agent Skills › QuixiAI/Hexis › connector-action-authorization

connector-action-authorization

GitHub

用于管理外部连接器（如邮件、消息平台）的操作权限，包括检查、授予和撤销发送、回复、标签及垃圾邮件处理等策略。遵循最小权限原则，支持限制目标、频率及自主行为范围，确保操作安全可控。

skills/installed/connector-action-authorization/SKILL.md QuixiAI/Hexis

Trigger Scenarios

用户要求通过连接器发送邮件或回复消息用户需要授予或撤销连接器操作权限用户希望配置自动化的连接器行为策略

Install

npx skills add QuixiAI/Hexis --skill connector-action-authorization -g -y

More Options

Non-standard path

npx skills add https://github.com/QuixiAI/Hexis/tree/main/skills/installed/connector-action-authorization -g -y

Use without installing

npx skills use QuixiAI/Hexis@connector-action-authorization

指定 Agent (Claude Code)

npx skills add QuixiAI/Hexis --skill connector-action-authorization -a claude-code -g -y

安装 repo 全部 skill

npx skills add QuixiAI/Hexis --all -g -y

预览 repo 内 skill

npx skills add QuixiAI/Hexis --list

SKILL.md

Frontmatter

{
    "name": "connector-action-authorization",
    "category": "communication",
    "contexts": [
        "chat"
    ],
    "requires": {
        "tools": [
            "connector_action_policy_status",
            "grant_connector_action_policy",
            "revoke_connector_action_policy"
        ]
    },
    "bound_tools": [
        "connector_action_policy_status",
        "grant_connector_action_policy",
        "revoke_connector_action_policy"
    ],
    "description": "Inspect, grant, and revoke scoped connector action policies for sends, replies, labels, spam triage, and provider state changes"
}

Connector Action Authorization

Use this when the user wants Samantha to act through an external connector beyond read-only setup/backfill: sending messages, replying to email, marking messages read, labeling, spam triage, deletion, or cross-channel intervention.

Principles

Do not treat a connected account as permission to act. Connection grants access; action policies grant behavior.
Make the policy concrete before granting it: connector, action kind, target/account, contexts, limits, and expiration when relevant.
Prefer narrow constraints: allowed_targets / allowed_recipients and max_per_day are safer than broad grants.
allow_autonomous: true means heartbeat or another non-chat context may act without a live user message. Use it only when the user explicitly asks for ongoing/autonomous behavior.
Keep destructive actions, especially delete, out of autonomous policy unless the user is exceptionally explicit and the constraints are narrow.

Flow

Call connector_action_policy_status to inspect existing grants.
If granting, summarize the proposed scope in normal language before calling grant_connector_action_policy.
Use requires_per_action_approval: false only when the user asks for preauthorization rather than case-by-case approval.
Use revoke_connector_action_policy when the user asks to remove or narrow a grant.