← 返回 Skills 市场
92
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install meeting-assistant-claw
功能描述
全流程会议助理虾。核心职责:录音转写、提炼待办并分发任务。业务价值:结果闭环——直接将待办指派到人的协作软件中,防止烂尾。 激活场景:用户发送会议录音文件(mp3/m4a/opus/wav/flac 等音频格式),要求转写、记纪要、提炼待办、分发任务。 触发关键词:转写录音、会议纪要、会议记录、提炼待办、会议总结...
安全使用建议
Before installing or running this skill, consider the following:
- Metadata mismatch: The SKILL.md requires FEISHU_APP_ID/FEISHU_APP_SECRET and a local SenseVoice transcription environment, but the published registry entry lists no required env vars or installs. Ask the publisher to update metadata to list these requirements.
- Sensitive credentials: FEISHU_APP_ID/SECRET give the skill the ability to obtain tenant_access_token and read/send messages and access contact lists. Only provide credentials for a purpose-scoped app and review app permissions (im:message:send_as_bot, contact:contact.base:readonly, im:chat).
- Data exposure & privacy: Audio may contain sensitive information. Transcripts are written to disk (~./transcripts, ~/.cache/modelscope). Decide whether storing transcripts locally or sending notifications to Feishu is acceptable for your data policy.
- Trust the transcription stack: The skill expects to use SenseVoice/FunASR and will download models into ~/.cache/modelscope. Ensure these packages and model sources are trusted and obtained from official releases to avoid malicious binaries.
- Plugin dependency: It prefers an internal openclaw-lark plugin if present; otherwise it will call Feishu APIs directly. Confirm presence and integrity of that plugin if you intend to use it.
- Test safely: If you proceed, test in an isolated/dev account first (limited-permission Feishu app, non-production audio) to verify behavior and tokens are handled correctly.
If the publisher cannot or will not clearly declare required env vars, install steps, and the Feishu app scope, treat the skill as untrusted and avoid providing production credentials.
功能分析
Type: OpenClaw Skill
Name: meeting-assistant-claw
Version: 1.0.0
The skill is a legitimate meeting assistant designed to transcribe audio files, summarize content, and distribute tasks via Feishu (Lark). It uses the SenseVoice-Small model for local transcription and provides a clear, multi-step workflow in SKILL.md that includes a mandatory user confirmation step before sending any external messages. While it handles sensitive credentials (FEISHU_APP_ID/SECRET) and executes Python scripts for transcription, these actions are strictly aligned with its stated purpose and use official API endpoints (open.feishu.cn). No evidence of data exfiltration, malicious persistence, or prompt injection was found.
能力评估
Purpose & Capability
The stated purpose (transcribe audio, extract tasks, and send them via Feishu) aligns with the instructions. However, the SKILL.md demands a large local transcription stack (Python venv, funasr/modelscope/onnxruntime/torch, cached models) and Feishu app credentials, while the registry metadata lists no required env vars or binaries. That metadata omission is inconsistent and reduces transparency.
Instruction Scope
Runtime instructions perform local file I/O (reading audio, writing transcripts under ~/.cache and ./transcripts, checking ~/.openclaw/extensions), run a Python one-liner to invoke local models, and call Feishu APIs to read contacts and send messages. Those actions are within the skill's described purpose, but they reference other local components/paths (sensevoice-transcribe skill, openclaw-lark plugin) and require access to contact lists and networked Feishu endpoints — all of which should be explicitly declared.
Install Mechanism
There is no install spec (instruction-only), which is low-risk from an installer standpoint. But the skill mandates preinstalled heavy ML packages and model caches; those will download large model files to ~/.cache/modelscope and require CPU/GPU resources. Expect significant disk, network, and dependency footprints despite no install recipe being provided.
Credentials
SKILL.md requires FEISHU_APP_ID and FEISHU_APP_SECRET, plus access to the Feishu contact API (reading user open_id), yet the skill metadata lists no required environment variables or credentials. Requesting credentials that grant access to the org's user directory and sending messages is sensitive and should be declared explicitly in metadata and justified by the skill.
Persistence & Privilege
The skill is not marked always:true and does not request permanent platform-wide privileges. It can be invoked by the agent (normal default). It reads/writes files within user-visible paths and checks for an optional plugin; it does not appear to modify other skills or system-wide agent settings.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install meeting-assistant-claw - 安装完成后,直接呼叫该 Skill 的名称或使用
/meeting-assistant-claw触发 - 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
首次发布:录音转写→会议纪要→待办提取→飞书消息分发一站式闭环
元数据
常见问题
全流程会议助理虾 是什么?
全流程会议助理虾。核心职责:录音转写、提炼待办并分发任务。业务价值:结果闭环——直接将待办指派到人的协作软件中,防止烂尾。 激活场景:用户发送会议录音文件(mp3/m4a/opus/wav/flac 等音频格式),要求转写、记纪要、提炼待办、分发任务。 触发关键词:转写录音、会议纪要、会议记录、提炼待办、会议总结... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 92 次。
如何安装 全流程会议助理虾?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install meeting-assistant-claw」即可一键安装,无需额外配置。
全流程会议助理虾 是免费的吗?
是的,全流程会议助理虾 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
全流程会议助理虾 支持哪些平台?
全流程会议助理虾 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 全流程会议助理虾?
由 Ricky(@tujinsama)开发并维护,当前版本 v1.0.0。
推荐 Skills