@injectailabs/emdash-injectai

AI chatbot plugin for EmDash CMS — powered by your own content.

Upload PDFs, videos, and documents. Your visitors get an AI assistant that answers questions grounded in that content — no hallucinations, no generic responses.

npm install @injectailabs/emdash-injectai

---

Requirements

• EmDash CMS >=0.1.0
• Astro >=4.0.0 with output: "server"
• Node.js >=18.0.0
• An InjectAI Labs account — you need an API endpoint URL and a secret key

---

Installation

1. Install the package

npm install @injectailabs/emdash-injectai

2. Register in astro.config.mjs

import { defineConfig } from "astro/config";
import cloudflare from "@astrojs/cloudflare";
import react from "@astrojs/react";
import emdash from "emdash/astro";
import { injectAI } from "@injectailabs/emdash-injectai";

export default defineConfig({
  output: "server",
  adapter: cloudflare(),
  vite: {
    optimizeDeps: {
      exclude: [
        "@injectailabs/emdash-injectai",
        "@injectailabs/emdash-injectai/sandbox",
        "@injectailabs/emdash-injectai/admin",
        "@injectailabs/emdash-injectai/astro",
      ],
    },
    ssr: {
      external: [
        "@injectailabs/emdash-injectai",
        "@injectailabs/emdash-injectai/sandbox",
      ],
    },
  },
  integrations: [
    react(),
    emdash({
      plugins: [injectAI()],
    }),
  ],
});

The vite.optimizeDeps.exclude and vite.ssr.external entries are required. Without them, Vite's dep optimizer will try to pre-bundle the plugin and cause worker crashes in dev mode.

3. Start your dev server

npm run dev

Navigate to /_emdash/admin/plugins/injectai — the plugin panel will appear.

---

First-time setup

Step 1 — Connect your API

Go to Admin → Plugins → InjectAI → Settings.

Field	What to enter
API Endpoint	Your InjectAI Modal deployment URL, e.g. https://yourendpoint.injectailabs.space
Secret Key	Your iai_xxxx secret key — used for uploads and admin operations only
Top K Results	How many document chunks to retrieve per query (default: 5)
Bot Logo URL	Optional — a 40×40px PNG/SVG shown in the chat header

Click Test Connection to verify the endpoint is reachable, then Save Settings.

The secret key is stored server-side only. It is never sent to the browser or included in any public API response.

Step 2 — Configure your AI agent

Go to the Settings tab, scroll to Agent Configuration.

Field	Description
Assistant Name	What the bot calls itself (e.g. "Acme Support")
Brand / Company	The company it represents
Tone	Communication style (e.g. "helpful, concise, professional")
Custom Persona	Optional — overrides the auto-built system prompt entirely
Provider	AI provider: cerebras, groq, openai, anthropic, google
Model	Model name for the chosen provider
API Key	Provider API key — stored encrypted on the server
Temperature	0 = deterministic, 1 = creative. Recommended: 0.1 for RAG
Max Tokens	Max response length (100–8000)

Click Save Agent Config to Server. Use Load Current Config to pull back what's stored.

Step 3 — Upload content

Go to Admin → Plugins → InjectAI → Upload.

Drag and drop files or click to browse. Supported formats:

• Documents: PDF, DOCX, PPTX, TXT, XLSX, CSV
• Media: MP4, AVI, MOV, MP3, WAV
• Images: JPG, PNG

Files up to 5 GB are supported. After uploading, switch to the Files tab to monitor processing status.

Processing stages: uploading → processing → completed (or failed).

Step 4 — Design the chat widget

Go to Admin → Plugins → InjectAI → Chat Designer.

Customize:

• Mode — Floating button (bottom-right) or Full-page (ChatGPT-style with sidebar)
• Theme — Dark or Light
• Colors — Accent, background, user bubble, bot bubble
• Typography — Font size, border radius
• Content — Title, subtitle, placeholder text, welcome message, suggestion chips

A live preview updates as you change settings.

---

Embedding the chat widget

Option A — Specific page via content editor

Open any page in the EmDash content editor, type / and select InjectAI Chat from the block menu.

The widget renders using the mode set in Chat Designer — floating or full-page.

Option B — Astro template (manual)

Import the component directly in any .astro file:

---
import InjectAIChat from "@injectailabs/emdash-injectai/components";
---

<!-- Uses Chat Designer mode (floating or full-page) -->
<InjectAIChat />

<!-- Always floating, regardless of Chat Designer setting -->
<InjectAIChat forceFloating />

---

Admin panel tabs

Tab	What it does
Settings	API endpoint, secret key, agent config
Upload	Upload files to the knowledge base
Files	View uploaded files, check processing status, delete
Generate	Generate modules, quizzes, or checklists from uploaded content
Chat History	Browse and inspect all public chat sessions
Chat Designer	Customize widget appearance with live preview
Documentation	Quick reference

---

Generating content from uploads

Go to Admin → Plugins → InjectAI → Generate.

1. Choose a content type: Module (HTML), Quiz (JSON), or Checklist (JSON)
2. Enter the Upload ID or Batch ID from the Files tab
3. Optionally edit the system/user prompts
4. Click Generate

The provider and model are read from your saved Agent Configuration.

---

Troubleshooting

"API endpoint not configured" in the chat widget → Go to Settings, enter your Modal URL, and click Save.

"Secret key not configured" when saving agent config or uploading → Enter your secret key in Settings and save.

Site Token shows "(generating…)" → Save Settings once — this triggers token generation.

Files stuck in processing → Check your Modal deployment logs. Processing is handled asynchronously by the backend.

Chat widget not rendering on a specific page → Make sure the page renders its content through <PortableText>. If your template uses a custom type map, add "injectai-chat" to it:

import InjectAIChat from "@injectailabs/emdash-injectai/components";

const myTypes = {
  // ...your existing types
  "injectai-chat": InjectAIChat,
};

---

Support

• Email: injectailabs.space@gmail.com
• Website: injectailabs.space
• Issues: github.com/muzammildafedar/emdash-injectai/issues

---

License

MIT © InjectAI Labs