Fire Rabbit Matrix

Problem

Multi-platform content distribution often breaks down on account selection, platform-specific rules, asset paths, scheduling limits, and duplicate publish tracking. Fire Rabbit Matrix wraps the local client’s HTTP surface into a small API set that scripts, agents, or automation pipelines can call: use loadAccounts to find eligible accounts, publish image or video posts according to platform rules, and optionally pull unpublished cloud works, start nurturing tasks, or generate copy.

How It Works

The skill assumes the Fire Rabbit Matrix client is running and logged in. Endpoints are POST requests where the JSON body is passed directly as the method parameters. The main capabilities are:
- Account discovery: returns accounts and groups; callers usually filter by platform, offline, and partitionId.
- Content publishing: supports image and video, uses half-width # for hashtags, requires absolute asset paths, and allows up to 9 images per post.
- Cloud publishing: pulls unpublished works from the remote workspace, marks them published after completion, and supports keyword filtering, image-to-video conversion, and BGM paths.
- Nurturing tasks: performs browsing, liking, and bookmarking on Xiaohongshu, Douyin, and Kuaishou, with configurable browse counts, dwell times, and like rates.
- AI copy: generates posting copy when an AI Key has been configured.

Boundaries

Publishing is a long-running operation, typically 20 to 90 seconds, so HTTP clients should set timeout ≥ 180s. Do not issue concurrent publish calls for the same account, and do not start multiple nurturing tasks at once. The service is bound to 127.0.0.1 without authentication; cross-machine access requires a reverse proxy and auth layer. Platform constraints matter: Xiaolüshu supports images only, nurturing is limited to Xiaohongshu, Douyin, and Kuaishou, and scheduling support varies by platform.