跳到主内容
智客 ZICQ

技能库 智客分类:Agent 工作流 agent-slack

Agent Slack

代理服务器的clack CLI:读取 URL/threads/history/unreads/later/canvases/workflows,从Markdown创建/编辑画布,搜索消息/文件,下载附件,搜索用户,列表/创建/邀请频道,打开DM,编写信件,管理Slack-native草稿,调度发送,以及清晰 发送/编辑/删除/反应/标记读取/上载.

3626 安装量

官方网址:skills.sh

技能介绍

先看中文介绍;官方 description 原文单独保留,不改写 SKILL.md。

做什么

代理服务器的clack CLI:读取 URL/threads/history/unreads/later/canvases/workflows,从Markdown创建/编辑画布,搜索消息/文件,下载附件,搜索用户,列表/创建/邀请频道,打开DM,编写信件,管理Slack-native草稿,调度发送,以及清晰 发送/编辑/删除/反应/标记读取/上载.

何时用

官方 description 未单独写出 Use when。按规范,代理会在用户任务与这段 description 的关键词匹配时激活本技能。

代理如何加载

按 Agent Skills 渐进披露:启动时只加载 name 与 description(约 100 token);任务匹配后才读入整份 SKILL.md 正文;scripts/、references/、assets/ 仅在需要时再读。 本文件正文结构:agent-slack、Safety、Workflow、Conditional references。 其中含规范建议的小节:分步指令。

文件分析

文件分析:除 SKILL.md 外,正文引用了 references/targets.md、references/output.md,属于带资源的技能包,这些文件按需再读。

官方 description(原文)

Slack CLI for agents: read URLs/threads/history/unreads/later/canvases/workflows, create/edit canvases from Markdown, search messages/files, download attachments, lookup users, list/create/invite channels, open DMs, compose messages, manage Slack-native drafts, schedule sends, and explicit sends/edits/deletes/reactions/mark-read/uploads.

agent-slackSafetyWorkflowConditional references

来源分类:skills.sh agent-skill

SKILL.md 与 Agent 调用

官方规范 ↗
name
agent-slack
description
Slack CLI for agents: read URLs/threads/history/unreads/later/canvases/workflows, create/edit canvases from Markdown, search messages/files, download attachments, lookup users, list/create/invite channels, open DMs, compose messages, manage Slack-native drafts, schedule sends, and explicit sends/edits/deletes/reactions/mark-read/uploads.
  1. 发现技能客户端向 Agent 提供名称与描述目录。
  2. 匹配与调用用户指定或任务匹配后,载入 SKILL.md 指令。
  3. 按需加载按步骤读取参考文档、使用脚本与素材。
指令中引用的文件 · 2
  • references/targets.md
  • references/output.md

以下路径提取自原文;文件是否齐全请以来源仓库中的完整目录为准。

具体调用语法与可用工具以目标 Agent 客户端为准。 查看调用机制说明 ↗

安装这个技能

Skills CLI ↗

先选择目标 Agent 和安装范围,保留技能包的附属文件,安装后检查客户端能否发现该技能。

该技能引用了附属文件,请从来源获取完整目录;仅复制 SKILL.md 可能缺少依赖。

交给 Agent 安装

复制安装指令给支持 Agent Skills 的代理,确认其中的目标目录与客户端匹配。

把 Agent Skill「agent-slack」安装到我的项目:SKILL.md 原文与官方 description 见 https://zicq.com/zh/skills/skl-0ecd49fcba456f76-%E5%8F%B2%E8%8E%B1%E5%85%8B%E6%8E%A2%E5%91%98.html
请存为 .cursor/skills/agent-slack/SKILL.md 或 .claude/skills/agent-slack/SKILL.md,frontmatter 的 name 与 description 保持原样,不要改写。
该技能还带 scripts/、references/、assets/ 等文件,请从 https://github.com/stablyai/agent-slack 取完整目录,不要只建一个 SKILL.md。

GitHub 完整包 ↗

终端安装 · Skills CLI

需要 Node.js 与 npx。先查看仓库技能列表,确认实际名称。

npx skills add 'https://github.com/stablyai/agent-slack' --list

npx skills add 'https://github.com/stablyai/agent-slack' --skill 'agent-slack'

CLI 会交互选择目标 Agent,默认安装到项目;用户级安装使用 -g。先通过查看命令核对仓库内容,再用 npx skills list 检查已安装技能。

阅读排版
--- name: agent-slack description: "Slack CLI for agents: read URLs/threads/history/unreads/later/canvases/workflows, create/edit canvases from Markdown, search messages/files, download attachments, lookup users, list/create/invite channels, open DMs, compose messages, manage Slack-native drafts, schedule sends, and explicit sends/edits/deletes/reactions/mark-read/uploads." --- # agent-slack Use `agent-slack` from `$PATH`. If it is missing, install it with: ```bash curl -fsSL https://raw.githubusercontent.com/stablyai/agent-slack/main/install.sh | sh ``` Fallback: `npm i -g agent-slack` (Node >= 22.5). Run `agent-slack --help` or the relevant subcommand help before guessing a command or flag. If a capability named here is absent from installed help, report version skew instead of guessing. Do not self-update the CLI without explicit authorization. ## Safety - Read and search freely. - Perform write actions only when explicitly requested: sends, edits, deletes, reactions, invitations, channel or canvas creation/editing, mark-read operations, scheduling or canceling delivery, uploads, Later state/reminder changes, DM/group-DM creation, and `workflow run`. Workflow runs can execute downstream actions. - For compose- or review-only requests, return proposed text without invoking Slack, or use `message draft create` to add a Slack-native draft the user can review and send (nothing is posted). `message compose` is send-capable; use it only when the user explicitly asks to open the interactive editor. In CI or another noninteractive environment, do not invoke it without separate authorization to send immediately: CI skips the editor and sends supplied text. - With `AGENT_SLACK_SAFE_MODE=1` (or the global `--safe-mode` flag) set, safe mode is enforced at the tool level: `message send` is redirected to the draft editor and `message edit`/`message delete` are blocked. Use it when nothing should post without human review. ## Workflow 1. Run `agent-slack auth whoami`. If needed, import credentials with `auth import-desktop`, `auth import-brave`, `auth import-chrome`, or `auth import-firefox`, then run `auth test`. 2. Prefer a Slack message URL when one is available. It carries the workspace, channel, and timestamp needed by most message operations. 3. Choose the narrowest read operation: `message get` for one message, `message list` for a full thread or channel history, and `search messages` or `search files` for discovery. 4. Use output limits such as `--limit`, `--max-body-chars`, and `--max-content-chars` to avoid unnecessary context. 5. For a requested write, execute only the requested mutation and verify the resulting JSON metadata. For scheduled writes, prefer `--schedule` with an ISO 8601 timestamp and explicit offset when timezone matters. Named `--schedule-in` phrases use the executing environment's local timezone; confirm that it matches the user's intent. Named `later remind --in` values such as `tomorrow` or `monday` also use the executing environment's local timezone at 9:00. Confirm that timezone or pass an explicit Unix timestamp. Use `--no-unfurl` with `message send` or `message compose` when the user wants Slack link and media previews suppressed. It cannot be combined with `message send --attach`. Ordinary `message send` and `message edit` calls auto-convert lists. `message send --blocks` and `message edit --blocks` use supplied Block Kit blocks, while `message send --attach` sends its initial comment without automatic list conversion. Inside auto-converted lists, use Slack's `` syntax because CommonMark `[label](URL)` links are not converted into labeled link elements. Slack-native drafts (`message draft list|create|update|delete`) manage drafts that appear in the user's Slack client; `create` posts nothing. `create` and `update` accept repeatable `--attach `; on `update` the files are added to the draft's existing attachments rather than replacing them. They use undocumented session endpoints and require browser-style auth (xoxc/xoxd). `canvas edit` uses Slack's public `canvases.edit` API and applies exactly one operation per call. The default `replace` operation replaces the whole canvas; section-targeted inserts/replacements and deletes require the section ID returned by Slack's Canvas tooling, while `rename` takes `--title`. Content operations take exactly one `--file` or `--markdown` source. It requires a standard token with `canvases:write`; imported browser credentials can create standalone canvases but cannot edit. ## Conditional references - Read [references/targets.md](references/targets.md) only when choosing between a message URL, channel, or user target, or when resolving multiple workspaces. - Read [references/output.md](references/output.md) only when handling returned message or canvas metadata, resolved users, or downloaded and failed attachments.

相关技能

Agent 工作流

Skill Creator

创造有效技能指南。 当用户想创造出新的技能(或更新现有的技能),以专业知识,工作流程,或工具集成来扩展克洛德的能力时,应该使用这种技能.

Agent 工作流

Clawdhub

使用ClawdHub CLI搜索,安装,更新并发布从taladhub.com的代理技能. 需要获取苍蝇上的新技能时使用,将安装的技能同步到最新版本或特定版本,或者发布 npm-instainddhub CLI 的新/更新的技能文件夹.

Agent 工作流

Agent Team Orchestration

管弦乐团多代理团队,任务设定周期,交接协议,审查工作流程. 使用时间: (1)建立2+特派员队伍,具有不同专业,(2)确定任务路线和生命周期(收录框_ spec_建设_审查_完成),(3)在特派员之间制定交接协议,(4)建立审查和质量关口,(5)管理特派员之间的交流和文物共享.

Agent 工作流

Superpowers

Spec-first,TDD,子代理驱动的软件开发工作流程. 当:(1)构建任何新功能或应用——触发脑暴_计划_子代理执行回路,(2)调试出一个bug或测试失败——触发系统性的根起过程,(3)用户说"让我们构建","帮助我计划","我想添加X",或"这个被打破",(4)完成一个功…