Integrations

The platform's integration philosophy is the same as its model philosophy: external services are capabilities to plug in, never dependencies to need. Every integration on this page is optional, configured in one place, and used by every agent once present. This page covers what each one adds and where it connects.

Levity: Image Generation

Visual content comes from Levity, a free companion project by the same developer: a local web studio for AI image generation that runs on the Cloudflare Workers AI free tier or on OpenRouter, with editing tools and export packs built in. Install it from the repository releases alongside the platform, and it serves both audiences, you get a full studio UI for hands-on image work, and the agents get a generation backend for theirs.

The connection is pleasantly self-organizing: agents learn your running Levity instance and save how to use it to the memory bank, after which the social media preset can attach generated images to posts and any agent can produce a visual when one serves the work. The image guidelines file in memory/system gives you art direction over all of it.

Adaptive Recall: Premium Memory

The bundled local memory backend works the moment the platform starts. Adaptive Recall is the upgrade path: knowledge graphs that connect related memories, multi-strategy retrieval, automatic consolidation, and ML-powered ranking that learns which memories help. Switching is a settings change, memory_backend to adaptive_recall plus your URL and token, with a free tier to evaluate, and every agent and tool works identically on either backend. The memory bank guide covers what the upgrade changes in practice.

Google Analytics and Search Console

Installs that run websites get measurement built in. Configure ga4_property_id and ga4_key_file for Analytics and gsc_site_url for Search Console in config/webpages.json, and the reporting tools pull traffic and search performance on demand, while the schedule node pulls reports hourly and cross-references them as evidence of what is working. The practical effect: when you ask the master agent how the site is doing, it answers from this week's actual numbers, and the learning layer correlates published work with measured outcomes.

BrightData: Structured Web Data

For high-volume structured collection, the BrightData integration wraps their scraping APIs: LinkedIn, Instagram, X, Facebook, TikTok, YouTube, Reddit, Google search results, and AI visibility tracking, how AI systems see and cite you. It connects with three values in settings, the API token, a SERP zone, and an unlocker zone, on Docker installs settable straight from .env. BrightData and the web browser bot split the scraping world cleanly: bulk structured data through the API, bespoke interactive work through the browser.

Email, SMS, and Messaging Providers

The outward-facing presets connect to providers you choose. The marketing preset takes email and SMS providers in config/marketing.json; the customer service preset takes IMAP and SMTP in config/customerservice.json, any standard mail account works; the chatbot takes platform tokens in config/chatbot.json; and the social media preset takes per-platform credentials in config/socialmedia.json. Every one of these is edited with validation in the Config tab, and the master agent's onboarding can walk each connection end to end.

Website Deployment

The webpages tool turns agents into web publishers: it builds and deploys sites to S3 or to API backends, configured in config/webpages.json with the optional api_url backend in settings for broadcast features. Combined with the coding pipeline for building and the Google reporting for measuring, an install can carry a site from creation through publication to performance review with the owner supervising rather than executing.