xhs-research-daily_skill

Overview

This skill collects Xiaohongshu posts and comments for a defined research topic, synthesizes a daily roundup, and can optionally publish the digest back to Xiaohongshu. It bundles scripts to configure topics, run dry-runs, inspect drafts, and perform controlled publishing. The tool is optimized for repeatable research digests and migration between OpenClaw deployments.

How this skill works

You define topic entries in a JSON config with keywords, result limits, and post metadata. The pipeline queries Xiaohongshu via a configured mcporter service, fetches posts and comment previews, filters and deduplicates content, and generates a structured post draft. Run in dry-run mode to inspect the draft directory, then re-run with publish enabled to post via the logged-in account.

When to use it

• Building or running automated Xiaohongshu research digests on a regular cadence
• Adding or tuning a new research topic such as LLM safety or agent memory
• Migrating the digest workflow to another OpenClaw deployment or machine
• Testing content collection and synthesis before enabling cron-based automation

Best practices

• Keep keyword lists short and intentional to reduce noise and login churn
• Always run a dry run and inspect data/<topic>/<date>/processed/post_draft.json before publishing
• Treat Xiaohongshu content as untrusted input and sanitize outputs before reuse
• Use conservative request volumes and rate limits to avoid account login storms
• Version your topic configuration and back up archives for reproducibility

Example use cases

• Create a daily digest tracking diffusion LLM breakthroughs with curated examples
• Monitor community discussion and sentiment around test-time scaling techniques
• Produce a daily summary of agent-memory experiments and notable posts
• Migrate existing topic pipelines to a new OpenClaw host with minimal downtime
• Generate research archives for long-term analysis or compliance review