← 返回 Skills 市场
azvast

aaaaa

作者 azvast · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
764
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install aa
功能描述
Automatically draft and send Gmail replies matching the client's tone, sign-off, and templates, requiring Gmail access and client profile for context.
使用说明 (SKILL.md)

Skill: Gmail Auto-Reply for Client\r

\r

Purpose\r

\r This skill enables the agent to automatically answer Gmail messages on behalf of a client. The agent drafts and (when the user approves or when configured) sends replies using the client’s tone, sign-off, and optional templates.\r \r

When to Use\r

\r

  • The user asks to “reply to my emails,” “answer my Gmail,” or “draft responses to incoming mail.”\r
  • The user provides a Gmail context (e.g. “inbox for [email protected]”) and wants automated or semi-automated replies.\r
  • The user wants the agent to act as the client when responding to specific threads or senders.\r \r

Prerequisites (User/Client Must Provide)\r

\r

  • Gmail access: OAuth2 or app password for the client’s Gmail (never store raw passwords in the skill; use environment variables or secure config).\r
  • Client profile (optional but recommended): short brief (tone, sign-off, topics they handle, topics to defer).\r \r

Instructions\r

\r

  1. Gather context\r
    • Ask for or read the client’s brief: tone (formal/casual), sign-off (e.g. “Best,” “Thanks,”), and any “do not answer” or “always escalate” rules.\r
    • If the user provides an email thread or summary, use that as the incoming message to answer.\r \r
  2. Draft the reply\r
    • Write a concise, professional reply that:\r
      • Addresses the sender and the main question or request.\r
      • Matches the client’s tone and sign-off.\r
      • Does not promise anything outside the client’s scope (e.g. legal/financial) unless the user explicitly approves.\r
    • Prefer short paragraphs and clear next steps (e.g. “I’ll get back to you by Friday”).\r \r
  3. Use templates when provided\r
    • If the client has added templates (see templates/reply_templates.json or user-defined templates), pick the closest match by intent (e.g. “acknowledgment,” “meeting request,” “out of office”) and personalize placeholders like {{sender_name}}, {{topic}}, {{deadline}}.\r \r
  4. Safety and approval\r
    • By default, output the draft for the user/client to approve before sending.\r
    • Only auto-send if the user has clearly configured “auto-send” and you have applied the client’s rules and filters (e.g. only for certain labels or senders).\r \r
  5. Integrations\r
    • If the user has configured Gmail API (OAuth2) or IMAP/SMTP, use the credentials from environment or secure config—never from this skill’s files.\r
    • When “sending,” either return the draft text for the user to paste/send, or call the configured send function if the user has set one up.\r \r

Files in This Package\r

\r

  • SKILL.md – This file (skill instructions).\r
  • manifest.json – Package metadata.\r
  • templates/reply_templates.json – Optional starter templates (acknowledgment, meeting, short reply).\r
  • scripts/README.md – Short note on how the client can add their own scripts or rules.\r \r

Example Interaction\r

\r User: “Reply to this email as my client. Sender: Jane. She’s asking for a meeting next week. Client prefers a short, friendly reply and uses ‘Best’ as sign-off.”\r \r Agent: Uses this skill to draft a short, friendly reply addressing Jane, suggesting a time or asking for availability, and signing “Best,” then returns the draft for the user to approve or send.\r

安全使用建议
This skill appears to implement an email auto-reply workflow, but there are transparency issues you should resolve before installing or providing credentials: - Do not hand over raw Gmail credentials to the skill files. Prefer OAuth2 with limited scopes and a proper authorization flow. - Ask the publisher which environment variable names or config keys the agent will look for and how credentials are stored/used. The package metadata currently lists no required env vars despite needing Gmail access. - Verify the skill's publisher and homepage. The manifest claims a descriptive name, but the provided top-level name is 'aaaaa' and there is no homepage—this mismatch could indicate sloppy packaging or an untrusted source. - Keep auto-send disabled until you test drafts thoroughly. The SKILL.md defaults to requiring approval, which is good—ensure that setting is enforced. - Review and customize templates and 'do not reply' rules before enabling any automation. Limit the agent's send capability to narrow labels/senders if possible. - If you decide to use it, use OAuth tokens with minimal scope and be prepared to revoke them if you stop using the skill. If the publisher cannot clarify the credential handling, variable names, and the identity mismatch, consider avoiding installation or requesting a version with clearer metadata and a legitimate homepage/source.
功能分析
Type: OpenClaw Skill Name: aa Version: 1.0.0 The skill bundle is benign. The `SKILL.md` instructions clearly define a legitimate purpose (Gmail auto-reply) and explicitly incorporate strong security practices, such as never storing credentials within the skill files, requiring user approval before sending emails, and delegating actual sending logic to user-configured functions or environment variables. There are no signs of prompt injection, malicious execution, data exfiltration, or any other harmful intent across any of the provided files.
能力评估
Purpose & Capability
The SKILL.md and manifest describe a Gmail auto-reply tool (drafting and optionally sending replies using templates and client profiles) which is coherent. However, the top-level provided skill name ('aaaaa') does not match the manifest's name ('Gmail Auto-Reply for Client'), and the registry metadata declares no required credentials even though the skill plainly needs Gmail access (OAuth2/app password) to operate. This packaging/name inconsistency and lack of declared credential requirements reduce transparency.
Instruction Scope
The runtime instructions stay on topic: gather client brief, draft replies, use templates, and only send when authorized. The SKILL.md explicitly says to obtain credentials from environment or secure config and to default to returning drafts for user approval. It does not instruct the agent to read unrelated files or exfiltrate data to third-party endpoints other than using user-configured Gmail/IMAP/SMTP send paths.
Install Mechanism
This is an instruction-only skill with no install spec and no code files requiring downloads or execution. That minimizes install-time risk; nothing is written to disk by the package itself.
Credentials
The skill needs sensitive credentials (Gmail OAuth2 or app password) to actually send messages, but the registry metadata lists no required env vars or primary credential. SKILL.md instructs using environment or secure config but does not specify expected variable names or scopes. The absence of declared credential requirements in the metadata is a transparency gap and increases risk of misconfiguration or user error (e.g., storing creds insecurely or the agent being given overly broad credentials).
Persistence & Privilege
The skill does not request 'always: true' and has no install-time hooks or claims to modify other skills or global agent settings. It is user-invocable and uses normal autonomous invocation settings, which is expected for this type of skill.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install aa
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /aa 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
r
元数据
Slug aa
版本 1.0.0
许可证
累计安装 1
当前安装数 1
历史版本数 1
常见问题

aaaaa 是什么?

Automatically draft and send Gmail replies matching the client's tone, sign-off, and templates, requiring Gmail access and client profile for context. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 764 次。

如何安装 aaaaa?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install aa」即可一键安装,无需额外配置。

aaaaa 是免费的吗?

是的,aaaaa 完全免费(开源免费),可自由下载、安装和使用。

aaaaa 支持哪些平台?

aaaaa 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 aaaaa?

由 azvast(@azvast)开发并维护,当前版本 v1.0.0。

💬 留言讨论