The Dashboard Is Dead. The Skill Replaced It.
A year ago, when an agency wanted to know what was being said about a topic online, it opened five tabs: Twitter, Reddit, YouTube, Google, and perhaps a social listening tool like Brandwatch or Mention. The process took 30 minutes to an hour for a rarely complete result.
Today, a Claude Code plugin named last30days-skill does this work in 60 seconds. It aggregates discussions from Reddit, X/Twitter, YouTube, TikTok, Instagram, Hacker News, Polymarket, Bluesky, Truth Social, and the open web, produces a structured report with citations, and generates ready-to-use prompts tailored to the user's search intent. The project, created by Matt Van Horn, reached 2,800 GitHub stars and the top trending repo spot in a single day.
This is not just another tool. It is a symptom of a fundamental change in how market intelligence is produced and consumed. Static dashboards, weekly reports, and monitoring tools with charts are giving way to agent skills that execute research, analysis, and synthesis in a single command.
How last30days-skill Works
last30days-skill is distributed as a "skill" (plugin) for Claude Code and compatible environments. Its operation follows a precise multi-step workflow.
The initial step is intent parsing. The skill analyzes the user's query to identify the topic, target tool, and search type (recommendations, news, comparison, prompting). This classification determines the output format and priority sources.
The next step is multi-source collection. The skill queries over 10 platforms in parallel via a set of API keys and access mechanisms: ScrapeCreators for Reddit/TikTok/Instagram, third-party APIs for X/Twitter, YouTube, and the web. The skill declares permissions for Bash, Read, Write, AskUserQuestion, and WebSearch, allowing it to run scripts, read and write files, and perform web searches.
The final step is synthesis and scoring. Results are weighted according to an engagement-weighted scoring formula: relevance at 45%, recency at 25%, engagement at 30%. The produced report includes citations from each source, engagement statistics, and ready-to-use prompts for the user's target workflow.
Benchmarks published in release notes show execution times of 66 to 77 seconds with 65 to 114 threads, and concrete results such as 31 X posts with 191 likes, 20 YouTube videos with 685,000 views, and 10 web pages for a single query on "Seedance 2.0 access."
Why Agent Skills Are Fundamentally Different From Dashboards
The difference between a social listening dashboard and an agent skill is not a difference of degree. It is a difference in kind that deserves explanation, because it has profound implications for how agencies operate.
A traditional dashboard (Brandwatch, Mention, Sprout Social) is a passive monitoring tool. It collects data continuously, stores it, and displays it as charts and metrics. You must consult it, interpret the data, and draw your own conclusions. The dashboard answers "what is happening?" but does not answer "what should I do?"
An agent skill is an active research tool. It executes a specific mission on demand, in real time, with a user-defined objective. It does not merely collect data: it analyzes, prioritizes, synthesizes, and produces actionable recommendations. The skill directly answers "what should I do?" while taking the user's context into account.
For an agency, this distinction has concrete operational consequences. A dashboard requires a monthly subscription, learning time, regular consultation time, and an analyst to interpret data. An agent skill requires an API key, a command, and 60 seconds of waiting. The marginal cost of each search is nearly zero compared to the fixed cost of a SaaS subscription.
Dave Morin, a well-known tech entrepreneur, uses last30days daily and has publicly recommended it for pre-sales research, product monitoring, and competitive tracking.
Concrete Use Cases for Agencies
For a digital agency, the applications of last30days-skill (and more broadly, agent research skills) are immediate.
The first use case is client meeting preparation. Before a sales call or steering committee, run "/last30days [client name]" to get a summary of what has been publicly said about the client over the past month. New products launched, issues mentioned on Reddit, CEO tweets, press articles: everything is consolidated into a 2-minute report.
The second use case is rapid competitive intelligence. When a client asks "what are our competitors doing right now?", the answer no longer requires a week of junior work. An agent skill can produce a real-time comparative snapshot, with sources and engagement as relevance indicators.
The third use case is technology evaluation. A tweet from Dobrenkz shows using last30days to compare "Vercel's agent-browser and Playwright," instantly obtaining real community feedback rather than the marketing pages of both projects.
The fourth use case is trend tracking for content proposals. For content marketing agencies, knowing what is buzzing in real time is the difference between proposing relevant topics and proposing stale ones. A last30days scan before each editorial planning session ensures proposed topics are aligned with current discussions.
Limitations and Risks to Know
last30days-skill is not without flaws, and a serious agency must understand them before integrating it into workflows.
The first risk is dependency on third-party APIs. The skill relies on ScrapeCreators, X/Twitter keys, and other data providers. If a service goes down or changes its terms, the skill stops working for that source. This is an inherent fragility of any tool that aggregates unofficial sources.
The second risk is compliance. Accessing certain platforms via session tokens (AUTH_TOKEN and CT0 for X/Twitter) raises questions about compliance with those platforms' terms of service. Agencies working with clients in regulated sectors must assess this risk.
The third risk is synthesis quality. The skill produces a report, but report quality depends on collected data quality. On a niche topic with few discussions, the report will be thin. On a controversial topic, it might overrepresent extreme opinions that generate the most engagement.
The fourth risk is credential security. The skill requires API keys and authentication tokens stored in environment variables. Secure management of these credentials is the user's responsibility, and agencies must ensure these tokens are not exposed in Git repositories or shared environments.
What This Means for the Future of Agency Tools
last30days-skill is a symptom of a broader movement. Monolithic SaaS tools with complex graphical interfaces are gradually giving way to lightweight, composable skills that execute within the user's work context. Instead of leaving your terminal to open a dashboard, you run a command and get results directly in your workflow.
For agencies, this heralds a transformation in required competencies. Value will no longer lie in the ability to manipulate complex tools (Brandwatch, Semrush, Ahrefs) but in the ability to orchestrate agent skills to produce market intelligence quickly and in a targeted manner.
The Claude Code skills ecosystem is still young. last30days-skill is at version 2.9.5, with near-weekly releases and a rapidly growing community. Other research, monitoring, and analysis skills emerge every week on marketplaces like AgentSkills, Smithery, and ClawHub.
Agencies adopting these tools today gain a measurable competitive advantage: faster research, lower costs, and a reactivity that manual processes cannot match. Those who wait risk finding themselves with obsolete processes in a market where execution speed is becoming a key differentiator.
An often-overlooked dimension is the impact on junior team members and non-technical profiles. An account manager who can run /last30days before a client meeting produces better preparation than an analyst spending two hours on Google and Twitter. The skill democratizes access to market intelligence within the agency, reducing dependency on senior profiles for research and monitoring tasks.
The architecture of last30days-skill itself illustrates a deeper trend: the convergence between development tools and business intelligence tools. A skill that runs in a developer's terminal but produces reports usable by a sales director blurs the boundaries between departments. Agencies that understand this convergence and train their teams to use these tools cross-functionally will be best positioned to capture the value of the agentic revolution.
The practical takeaway for agency leaders is clear. Identify the three or four research workflows that consume the most time in your organization, whether that is client research, competitive analysis, trend monitoring, or technology evaluation. Then test whether an agent skill can replicate or accelerate each workflow. The results will likely vary: some workflows will see dramatic improvement, others will require human judgment that no skill can replace. But establishing this map of automatable versus non-automatable research tasks is the first step toward building an agency that operates at the speed the market now demands.
The distribution model itself is worth noting. last30days-skill is available through multiple skill marketplaces including AgentSkills, Smithery, and the ClawHub/OpenClaw ecosystem. This multi-marketplace distribution mirrors how browser extensions or mobile apps spread across platforms, and it suggests that the agent skill ecosystem is developing its own distribution infrastructure. For agencies building their own internal tools and skills, understanding these distribution channels now will matter as the ecosystem matures.
Finally, the pricing model of the underlying data sources deserves attention. While last30days-skill itself is MIT-licensed and free, the APIs it depends on are not. ScrapeCreators offers 100 free credits for testing, but sustained usage requires a paid plan. Brave Search API, OpenRouter, and the various AI model APIs each have their own pricing. An agency budgeting for agent skill adoption should account for these data access costs, which are typically modest compared to the analyst time they replace but can accumulate at scale.
