[Extension] Andre AI Real-Time Responder – Engage Discussions via Webhooks

Andre Software

Hi everyone,

I've just rolled out Andre AI Real-Time Responder, an event-driven automation extension designed to eliminate dead threads and instantly drive user engagement by connecting your Flarum board with any external AI framework or custom API endpoint.

Unlike static AI integrations, this extension hooks directly into Flarum's discussion lifecycle. The moment a user spins up a new thread, it safely captures the payload and dispatches it over a secure gateway, waiting for your custom microservice to process and return a high-context, architectural, or technical reply under a designated bot identity.

🛡️ Why Use It?
Zero Performance Hit: Implements robust asynchronous error handling and structured timeouts. If your external endpoint drops, Flarum handles the exception silently without breaking the user experience.

Granular Control: No hardcoded endpoints. The administrative panel gives you absolute control over server mapping and dynamic behavioral directives.

Payload Verification: Outgoing requests carry structural headers including a custom verification token (X-Signature: AndreSoftwareHub) to ensure your middleware only processes verified requests originating from your forum core.

⚙️ Admin Panel Gateway Features
Through the integrated backend UI, you can fully manage:

Interactive Hook Gateway Target: Map the exact URL pointing to your AI router or serverless function.

Security Bearer Token: Set up API secrets to pass authorization seamlessly.

Automated Bot Profile Assignment: Assign the database User ID that the AI should use to publish the automated responses.

Behavioral Directive Template: Pass system instructions to instantly shape the persona, technical boundaries, and tone of the AI response.

💻 Open Source Repository
The extension architecture is fully modular and light on the core database schema. You can check the code, review the file system configuration, and contribute directly over on GitHub:

👉 andrexdd/flarum-ai-realtime-responder

Released under the MIT License. Built for high-performance communities.

Looking forward to your feedback and hearing how you map it to your own webhooks!

Cheers,

Andre Software

luceos

Andre Software I'm a bit confused about this announcement. Aside from it missing an actually published version/tag, according to the code you will need to fill in the value https://api.fikiral.net.tr/v1/stream-ai for the api endpoint, but there's no explanation in your readme, here in your op, or elsewhere on where to get tokens and whether there are any costs involved..

Also some necessary information is missing I believe, but a mod will probably point you to the requirements.

Andre Software

luceos Hey @luceos , thanks for taking the time to review the repository. Appreciate the solid feedback.

To clarify the confusion regarding the architecture:

The Endpoint & Cost: The URL https://api.fikiral.net.tr/v1/stream-ai shown in the setup documentation is strictly an example placeholder (mock server context) used during our internal alpha testing phase. This extension does NOT enforce any hardcoded third-party routing, nor does it require any paid subscriptions from our end. It is completely standalone. Ghee developers are supposed to spin up their own serverless functions, AI middleware, or microservices, and map their unique local/private endpoints right into the admin panel gateway.

Authentication Tokens: Since the architecture is designed to interface with whatever private backend infrastructure the forum administrator deploys, the 'Bearer Token' configuration is entirely managed by the forum owner's own server auth rules. There are no proprietary tokens to purchase from us.

Stable Release Tagging: You are totally right about the package discovery. We are currently polishing the stable build lifecycle and will be pushing the formal semantic version tags (v1.0.0) to GitHub later today so it registers seamlessly via Packagist/Composer.

I will update the README.md right away to explicitly state that the endpoint mapping is fully open-ended and infrastructure-agnostic to avoid any further misunderstanding. Thanks again for pointing it out!

Andre Software

luceos Quick update: The README.md has been successfully updated to clear up any deployment context ambiguity, and the repository is now formally tagged under the stable v1.0.0 release. Thanks again for the guidance, appreciate it!