mcpbeat

Building Slack Agents

vercel-labs/building slack agents

679 tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
9
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/vercel-labs/academy-skills --skill Building Slack Agents

The instruction itself

14 sections, as written by the author

Building Slack Agents

Companion skill for the Building Slack Agents course. Build and deploy a Slack agent on Vercel in a single session.

Commands

/slack-agents learn

Start the guided learning loop. Fetches lessons from Academy and drives you through the course.

/slack-agents new

Scaffold and deploy a new Slack agent. Five-stage wizard:

  • Project setup — scaffold from Slack Agent Template, configure LLM provider
  • Slack app creation — generate customized manifest.json, walk through Slack console
  • Environment configuration — signing secret, bot token, API keys with validation
  • Local testing — dev server + ngrok tunnel, test in Slack
  • Production deployment — deploy to Vercel with env vars configured

/slack-agents submit

Evaluate your current implementation against the active lesson's outcomes.

Content source

https://vercel.com/academy/slack-agents.md           → course overview
https://vercel.com/academy/slack-agents/<lesson>.md   → lesson content

Core concepts

Slack Agent Template

Starting point with OAuth, event subscriptions, and signature verification pre-configured.

Workflow DevKit

Durable agent execution: suspend mid-conversation, wait for human approval, resume. Tool calls auto-retry on failure. Responses stream to Slack in real time.

Human-in-the-loop

Built-in approval pattern: agent posts Approve/Reject buttons, workflow pauses (no compute), resumes on click.

AI Gateway

Route to any model via gateway('anthropic/claude-sonnet-4.6'). Automatic failovers during provider outages.

Tools

Custom tools connect to your systems: customer records, support tickets, databases. Each tool has a typed schema so the agent knows when and how to use it.

Progress detection

| Signal | Stage |

|---|---|

| Project scaffolded from template | 1: Setup |

| manifest.json with correct scopes | 2: Slack app |

| .env.local with SLACK_SIGNING_SECRET + SLACK_BOT_TOKEN | 3: Config |

| Dev server running, ngrok tunnel active | 4: Testing |

| Deployed to Vercel with env vars | 5: Production |

Teaching guidelines

  • This is a wizard-heavy skill — guide through external console steps (Slack API, Vercel dashboard)
  • Validate credentials before moving on (bad tokens waste time)
  • When opening external URLs, tell the user exactly what to click
  • Test the bot in Slack before deploying — catching issues locally is faster

How to use it

Copy the folder

Take vercel-labs/building slack agents from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.