getting-started — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited getting-started (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Economic Mind is a library of bottom-up market perspectives and hard company financials, served to Claude through a connector. This plugin adds skills that turn that data into rigorous, fully-sourced analysis - every figure traces back to its source, and the deliverable is data (Markdown + JSON) that the client is free to style. Your job when this skill runs is to orient the user: explain what's here, show how to trigger it, confirm the connector is connected, and point them to the right skill for their goal. Keep it brief and concrete - a tour, not a manual.
There are two layers, and it helps to keep them distinct:
https://mcp.economicmind.com/mcp) that exposes EconomicMind's data as tools: curated market reports, and multi-year company financials. Claude calls these tools automatically when a request needs data; the user does not call them by hand.
pipeline that pulls connector data, works to files, and assembles a fully-sourced deliverable.
margins and growth (each figure footnoted to its formula), plus web-sourced profile facts. Try: "give me Sperre AS at a glance", "pull the financials for <company>", "what do we know about <company>".
cross-company summary of size, growth and profitability, normalized to one currency. Try: "benchmark <A>, <B> and <C>", "build a comp table for the Nordic fire-protection market", "how does <company> stack up against its peers".
per-company build) and stress-test the headline against credible cross-checks. Try: "size the Nordic 3PL market bottom-up", "how big is <market> and how sure are we", "show the market-size buildup for <market>".
These are the data tools the skills (and Claude) draw on. Coverage is broad - market reports span many topics and geographies, and company financials run across multiple years, with the deepest depth in the Nordics.
size, scope and content catalog. "find market reports on autonomous trucking in the Nordics".
growth, …) in full. "open that market and show me the demand section".
"find the company Sperre AS".
a chosen currency, with each line attributed to its provider. "pull <company>'s revenue and EBITDA for the last three years in NOK".
| Way | How | Example |
|---|---|---|
| Natural language | Just describe what you want - works everywhere | "size the Nordic 3PL market bottom-up" |
| Skill by name | Invoke a skill directly | /economic-mind:company-at-a-glance |
| Connector tools | Called automatically by Claude when a request needs data | - (you don't call these yourself) |
Natural language is the primary path and the most reliable across surfaces. Invoking a skill by name is handy when you know exactly which one you want; the exact slash syntax depends on the surface (Claude Code vs. Claude.ai / Desktop).
Economic Mind is growing. Expect coverage to keep widening - more markets and geographies, and deeper company and report data - and new connector capabilities to arrive over time. The skills will track the connector as it grows, so the analysis you can ask for will broaden alongside the data.
After orienting the user, confirm the connector is actually connected before sending them off:
search_markets with a sample querythe user might care about - to confirm connectivity. Keep it to one call; do not probe exhaustively.
suggest a concrete next step: pick one of the three skills with a real prompt.
retry - in Claude Code via /mcp, or in Claude.ai / Desktop via the connector settings. Don't retry in a loop; once is enough to surface the prompt.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.