Weave Code
Filament Assistant Laravel Package
assistant-engine/filament-assistant
Integrates an AI assistant directly into Filament admin panels. Chat with an LLM, run tools, and automate workflows without leaving the dashboard. Configure providers and tools via simple PHP, with a clean chat UI built on Filament components. Ideal for Laravel teams adding in-panel AI help.
A Filament Assistant package that enables AI features, bringing advanced assistant capabilities directly into Filament.
Frequently asked questions about Filament Assistant
- How do I install assistant-engine/filament-assistant in a Laravel app?
- Install it with Composer using composer require assistant-engine/filament-assistant, then register the plugin in your Filament PanelProvider. Publish the package config and migrations, run the migrations, and add your model provider credentials. Most Filament AI plugins also need a queue worker for background or streaming work, so plan for that during setup.
- Which Laravel, PHP, and Filament versions does this package support?
- The package metadata does not publish a version matrix, so check its composer.json constraints before adopting it. Filament plugin APIs change across major releases, and low-traffic plugins often lag behind the newest Filament version. Pin your Filament and Laravel versions and confirm the supported combination with the maintainer or by reading the source.
- Do I need the assistant-engine core package as well?
- This package is a Filament presentation layer, not the AI engine itself, so it almost certainly depends on assistant-engine/laravel-assistant-engine for provider abstraction, message storage, and tool orchestration. Treat the core package as the real dependency decision. The Filament plugin adds panel registration, resources, and chat UI on top.
- How do I configure an LLM provider and API key?
- Provider settings are typically published into a config file and read from environment variables, though some packages support encrypted database storage. Check the published config to see which providers are supported and where credentials live. Never commit API keys, and restrict who can edit assistant settings in production.
- Does it support streaming responses, and do I need a queue worker?
- Streaming and tool execution usually run through queued jobs, so a working queue driver such as Redis or the database is likely required. Confirm whether the package uses broadcasting or a polling transport for the chat UI. Without a running queue worker, responses may stall or tools may not execute.
- Where are conversations and messages stored, and can I control retention?
- Assistant resources typically persist threads, messages, and tool payloads in your application database through published migrations. Review what is stored by default, because prompts and tool outputs can contain PII or regulated data. Add a retention or pruning strategy and confirm whether full message bodies are saved.
- Is the assistant multi-tenant aware in Filament panels?
- Multi-tenancy is not confirmed in the public metadata, so verify whether assistants and conversations are scoped to the current Filament tenant. If scoping is missing, you may need custom queries, policies, or a modified resource. This matters most when one admin panel serves multiple customers.
- Is it safe to give an admin-embedded assistant tool access in production?
- An assistant with tool access is a privileged actor inside your admin panel, so apply least privilege and allowlist only the tools it needs. Prompt injection through admin-editable fields is a real risk if tools can write data. Review the data flow to your LLM provider and log tool calls for audit.
- Can I use the AI assistant without Filament, or should I build my own UI?
- Yes. If your admin surface is not Filament, evaluate the headless core package directly and build the interface you need. Even in a Filament app, wrapping the core behind your own service interface keeps the UI swappable. This package buys you convenience, not engine functionality.
- How mature and actively maintained is filament-assistant?
- The package currently shows no stars or dependents, which means there is little external validation rather than proof of poor quality. Treat it as an unvetted, likely single-maintainer package and check its license, CI status, test coverage, and issue history. Pin versions and be prepared to maintain a fork if Filament releases breaking changes.
Weaver
How can I help you explore Laravel packages today?