Albye icon

Albye

Automate Albert security training in Slack using a local or cloud LLM.

A Chrome extension that watches your Albert DM, reads the conversation, and clicks the right buttons — with human-like delays so it feels natural.

Chrome extension Slack web Ollama · OpenAI · Anthropic · Gemini No Slack API token

How it works

  1. You open Albert's DM in Slack web.
  2. The extension reads recent messages and interactive elements (buttons, radios).
  3. An LLM decides: click a button, pick a radio, send text, or wait.
  4. The bot acts after a short random delay (default 3–6 seconds).

No Slack API token required. The extension uses your existing browser session on app.slack.com.

Requirements

What Notes
Google Chrome Or another Chromium browser (Edge, Brave, …)
Slack web app.slack.com — Albert app installed in your workspace
LLM Local: Ollama · Cloud: OpenAI, Anthropic, or Google Gemini API key

Option A — Local LLM (Ollama)

Recommended if you want everything to stay on your machine. Free after install.

1. Install Ollama

Download and install from ollama.com (macOS, Linux, or Windows).

2. Start the server

Ollama usually runs in the background after install. If needed:

ollama serve

3. Pull a model

Pick a model that handles JSON reasoning well. Example:

ollama pull gemma4:12b

Other options: llama3.2, qwen2.5, mistral — use whatever you already have; the popup lets you pick from installed models.

4. Verify Ollama is running

curl http://127.0.0.1:11434/api/tags

You should get JSON listing your installed models.

Default URL: http://127.0.0.1:11434
Change it in the extension popup only if Ollama runs on another host or port.

Install the extension

From source (today)

  1. Get the project folder — clone the repo or download a release zip. 
    git clone https://github.com/rbobillot/albye.git
cd albye
  2. Open chrome://extensions in Chrome.
  3. Enable Developer mode (top right).
  4. Click Load unpacked and select the albye folder.
  5. Pin the Albye icon to your toolbar.

Developers — build after clone

If you cloned the repo (not a release zip), bundle the content script once:

bun install
bun run build

Requires Bun. Re-run bun run build after editing content.js or lib/, then reload the extension and refresh Slack.

Chrome Web Store — publish with bun run pack (build + tests + dist/albye-<version>.zip). Use that zip for the store upload; link this page as homepage/support.
Install from source (today) — teammates use Load unpacked until the store listing is live.

After every extension update

  1. Click Reload on the extension card at chrome://extensions
  2. Refresh your Slack tab (F5) — required for the content script to reconnect

Optional — first-time setup

All settings live in the extension popup (saved to Chrome storage). Open Albert's DM in Slack, click the Albye icon, use Fill from current Slack tab, then Start (which tests the LLM automatically). Use Test connection first if you only want to verify the LLM without running.

Configure Slack

  1. Open Slack web and go to your Albert direct message.
  2. Click the extension icon → open the Slack panel.
  3. Click Fill from current Slack tab (easiest), or copy IDs from the URL:
https://app.slack.com/client/{WID}/{CID}
Field URL part
Workspace ID {WID} — first ID after /client/
Channel ID {CID} — Albert DM channel (usually starts with D)

When you're already on Albert's page, the Slack panel collapses automatically — use Open Albert tab in the header to jump back.

Configure the LLM

Open the extension popup → LLM panel.

Local (Ollama)

  1. Provider: Local (Ollama)
  2. Pick your model in the dropdown (or click Test connection to refresh the list)
  3. Start will test Ollama again before monitoring begins

Cloud (API key)

  1. Provider: OpenAI, Anthropic, or Google Gemini
  2. Paste your API key (stored locally on your device, never synced)
  3. Click Test connection to load models, or go straight to Start (tests the LLM first)
  4. Choose a model from the dropdown
Cloud providers: Albert conversation text is sent to the provider's API when the bot runs. Use Ollama if you need everything to stay on your machine.

Behavior (optional)

Start the bot

  1. Slack tab open on Albert's DM (same channel ID you configured).
  2. Extension popup → Start (tests LLM, then monitors).
  3. Status should show Running · on Albert channel. Activity logs refresh in the popup.

Debug in the Slack tab

Open DevTools (F12) → Console. You should see [Albye] logs.

__albye.dump()           // full diagnostic
__albye.getLlmPayload()  // payload sent to the LLM
__albye.forceTick()      // run one decision cycle

Privacy & data

Albye stores settings and optional API keys on your device only. With Ollama, course text stays on your machine; with cloud LLMs, conversation context is sent to the provider you choose.

Read the full privacy policy → (required for Chrome Web Store).

Troubleshooting

More detail in the repo: docs/TROUBLESHOOTING.md and docs/SETUP.md.

Setup checklist