Skip to content

last30days-skill

What it is

last30days-skill is a sophisticated AI agent skill for Claude Code, OpenClaw, and Gemini CLI. It acts as a specialized search and research engine that prioritizes real-time social signals (Reddit upvotes, X likes, YouTube transcripts, Polymarket odds) over traditional SEO-optimized web results, supporting the MCP 3.0 protocol.

What problem it solves

Traditional search engines often surface stale editorial content or SEO-spam. In the fast-moving AI ecosystem, critical information first appears in community discussions. /last30days bridges a dozen disconnected platforms, allowing an AI agent to search, score, and synthesize current trends, tool comparisons, and "unfiltered" community feedback from the last 30 days. It is a vital tool for agents using Claude 4.8 Opus and GPT-5.5 to stay current with weekly shifts.

Where it fits in the stack

Category: AI Assistants & Knowledge / Claude Code Skills. It functions as an MCP 3.0 server or native skill for development environments, integrating with Model Context Protocol (MCP) for dynamic resource retrieval.

Typical use cases

  • Deep Tool Comparison: Asking /last30days OpenClaw vs Hermes to see real-world performance reports and GitHub velocity instead of marketing pages.
  • Pre-Meeting Briefings: Quickly summarizing a person or company's activities over the last month (e.g., /last30days Peter Steinberger).
  • Trend Analysis: Understanding the latest best practices in prompt engineering or agentic workflows (e.g., /last30days Nano Banana Pro prompting).
  • Repository Onboarding: Summarizing the last 30 days of Git history and issues to get an agent up to speed on a new codebase.

Strengths

  • Social Scoring: Ranks information based on actual engagement (upvotes, engagement rates) rather than keyword density.
  • Parallel Search: Executes entity-aware subqueries across multiple platforms simultaneously, optimized for NVIDIA NIM inference.
  • Intelligent Pre-Research: The v3 engine resolves relevant handles, subreddits, and hashtags before searching, ensuring high-signal discovery.
  • Shareable Artifacts: Can emit self-contained, dark-mode HTML briefs for easy distribution in Slack or Notion.

Limitations

  • Token Usage: Parallel synthesis of multiple sources can consume significant input tokens if not carefully managed.
  • Rate Limits: Subject to the rate limits of the underlying search providers and social platforms.
  • Recency Bias: Explicitly ignores older, potentially more established documentation in favor of the "last 30 days" of activity.

When to use it

  • Emerging Tech Research: When researching tools or libraries that were released or updated very recently.
  • Vibe Checks: Understanding the community sentiment or "vibe" around a specific AI model or framework.
  • Crisis Monitoring: Tracking real-time outages, bugs, or major breaking changes reported by the community.

When not to use it

  • Deep Historical Research: If you need information from more than a month ago, traditional search is required.
  • Static Documentation: For stable libraries with unchanging APIs, official docs are more reliable than social chatter.
  • Critical Production Code: Social signals should not replace rigorous testing or official security advisories.

Getting started

Installation (Claude Code)

# Add the skill via the Claude Code plugin marketplace
/plugin marketplace add mvanhorn/last30days-skill

Installation (OpenClaw)

clawhub install last30days-official

Hello-World

# Research a topic
/last30days "Claude Code MCP servers"

CLI examples

Deep Tool Comparison with HTML Export

/last30days "OpenRouter vs DeepSeek" --emit=html --output=comparison.html

/last30days "vLLM PagedAttention bugs" --sources=github,reddit

Periodic Activity Summary

/last30days "Anthropic API updates" --frequency=weekly --summarize=bulleted

API examples

Integration via OpenClaw Skill API (Python)

from openclaw import SkillRunner

# Initialize the skill
skill = SkillRunner("last30days-skill")

# Execute a research query
brief = skill.execute(
    query="Llama 4 Maverick performance benchmarks",
    depth="detailed",
    platforms=["x", "reddit", "hacker-news"]
)

print(brief.summary)

Programmatic Webhook Trigger

// Trigger a last30days research task from an external event
fetch('http://localhost:3000/skills/last30days/run', {
  method: 'POST',
  body: JSON.stringify({ query: 'GPT-5.5 release rumors' })
});

Sources / references

Contribution Metadata

  • Last reviewed: 2026-06-28
  • Confidence: high