Approve AI Agent Skills from Slack — Human-in-the-Loop in Your Channel

Your AI agent runs a task, hits a manual checkpoint, and posts a message in your Slack channel asking for approval. Reply yes — the pipeline continues. Reply no — it stops. No dashboard, no browser tab, just Slack.

What You Will Build

A three-skill pipeline connected to your Slack workspace. The first two skills run automatically, then the pipeline pauses for your approval before the third runs:

At the manual step, your Slack bot posts a message and waits. You reply yes in the channel — the final skill runs and posts the result.

What You Need

• A Slack bot connected to OE Runtime — follow Connect OE Runtime to Slack first if you haven't already. Confirm that run hello-world responds in Slack before continuing here.
• Node.js 18+ for npx
• An API key from OpenAI, Anthropic, or any supported LLM provider

Create the Project Folder

Create a folder called chain-slack/. One root agent.yaml orchestrates all three skills — no separate YAML files per skill needed:

Create agent.yaml

The root agent.yaml lists all three skills in the order they run. auto skills fire immediately; manual skills pause for your approval in Slack.

name: Chain Slack — Approval Demo
description: Three-skill pipeline with automatic and manual approval stages.

skills:
  # Runs first — auto (no approval needed)
  - path: ./chain-test
    trigger_type: auto

  # Runs second — auto (fires immediately after chain-test)
  - path: ./chain-followup
    trigger_type: auto

  # Runs third — manual (pauses and asks for your approval in Slack)
  - path: ./chain-manual
    trigger_type: manual

Create the SKILL.md Files

---
name: chain-test
description: Greets you and chains to two follow-up skills. Tests auto and manual skill behaviour.
license: Apache-2.0
metadata:
  author: Open Enthrium
  version: "1.0"
---

You are a helpful assistant.

## Step 1: Greet
Say hello, mention today's date, and give one fun fact about today in history.

---
name: chain-followup
description: Runs automatically after the greeter and summarises its output.
license: Apache-2.0
metadata:
  author: Open Enthrium
  version: "1.0"
---

You are a summarizer assistant.

## Step 1: Summarize
The previous skill just ran. Summarize its output in one sentence
and say "Auto skill complete ✅".

---
name: chain-manual
description: Runs only after the user approves in Slack.
license: Apache-2.0
metadata:
  author: Open Enthrium
  version: "1.0"
---

You are an assistant that confirms tasks.

## Step 1: Confirm
The user approved running you from Slack.
Acknowledge the approval, briefly reference what the previous skills said,
and say "Manual approval complete ✅".

Create oe-config.json

Add your LLM key, server settings, and Slack bot token. Leave publicUrl blank — you will fill it in after the Cloudflare tunnel starts.

{
  "llm": {
    "provider": "openai",
    "model": "gpt-4o",
    "apiKey": "sk-..."
  },
  "server": {
    "enabled": true,
    "port": 3333,
    "publicUrl": "",
    "webhook": true
  },
  "connectors": [
    {
      "connection_name": "My Slack Bot",
      "connection_type": "slack",
      "botToken": "xoxb-YOUR-BOT-TOKEN",
      "auto_reply": true,
      "rate_limit": { "messages_per_minute": 10 },
      "history": { "enabled": true, "max_messages": 10 }
    }
  ]
}

Replace sk-... with your LLM API key and xoxb-YOUR-BOT-TOKEN with your Slack bot token. See Connect OE Runtime to Slack for how to create a Slack app and get the token.

Start a Cloudflare Tunnel

Slack needs a public HTTPS URL to deliver events to your server. Cloudflare Tunnel gives you one for free — no account or signup needed.

npx cloudflared tunnel --url http://localhost:3333

Copy that URL. Keep this terminal open — the tunnel must stay running while you test.

Update publicUrl and Start the Server

Open oe-config.json and paste the Cloudflare URL into publicUrl:

"publicUrl": "https://purple-mouse-1234.trycloudflare.com"

Open a second terminal and start OE Runtime from inside the chain-slack/ folder:

cd chain-slack
npx -y @openenthrium/oe-runtime@latest --serve --config oe-config.json

The server prints the exact webhook URL to paste into your Slack app. Make sure Socket Mode is OFF in your Slack app (Settings → Socket Mode) — when Socket Mode is on, Slack ignores the webhook URL and your server receives nothing.

Verify Event Subscriptions

In your Slack app at api.slack.com/apps, confirm that:

• Socket Mode is OFF
• Event Subscriptions → Request URL shows Verified ✓ with your current Cloudflare URL
• Subscribe to bot events includes message.channels and message.im
• The bot is invited to your channel: /invite @OE Bot

Test the Pipeline in Slack

In your Slack channel, type (no slash needed):

Reject the skill

Reply no instead of yes when the bot asks for approval:

Commands Reference

How It Works

1. Auto skills fire immediately — the output of each skill flows forward as context. All auto-skill results are bundled into the bot's reply in Slack.
2. Manual skills pause. The server stores the pending skill in memory and appends an approval prompt to the bot's message.
3. When you reply yes, the server runs the next skill with the previous output as context and posts the result to Slack.
4. When you reply no, the skill is discarded and the bot confirms cancellation.