@dr-xiaoming

kol-content-screening matches keyword content across chinese social platforms (douyin, xiaohongshu, toutiao, bilibili) using web search aggregation. outputs ranked, evidence-backed reports with explicit confidence and time-window framing, designed for pr and competitive intel workflows where platform apis are unavailable.

blueai-models covers configuration and selection of 100+ ai models (claude, gpt, gemini, deepseek, qwen, and others) via unified bluai relay, with quick-start scripts, image generation workflows, and model selection tables for openclaw integration.

blueai-nano-banana wraps gemini/nano-banana image generation via internal relay, outputting png locally. intended for rapid prototyping (moodboards, demo placeholders, social assets), not production design work.

bluehex-data-collector captures full conversation transcripts from openClaw agents every 2 hours, redacts credentials and pii, and rolls daily pull requests to a centralized monitoring repository for downstream ai-pm consumption.

catl-harness-pr covers github pr submission to a private catl knowledge-engineering repo, including fork vs legacy mode detection, preflight checks, multi-phase procedures, and (as of v0.4.0) daily auto-summarization with lark integration. targets bluefocus nanning project workflows only.

bluehex-data-monitoring captures openclaw conversations and media every 2 hours, redacts credentials and pii, writes sanitized markdown to local disk, and pushes rolling daily pull requests to a monitored repository. designed as raw data foundation for downstream ai-pm consumption.

蓝标场景下不同任务类型 → 推荐模型的实战路由策略，用于在主 session、子 agent、cron 任务里**节省 token 成本 + 加快执行速度**。覆盖 6 大任务类型（长文档/中文生成/编程 Agent/对话/分析/超低成本子任务），每条都给具体模型别名 + 真实失败案例 + 成本对比。触发：选模型...

Multi-platform social media data collection and aggregation for content performance tracking. Use when: (1) collecting engagement metrics (views/likes/commen...