How do you vet a skill or MCP server before your agent loads it?

Hi all, I'm Marius, and I built Spellwarden.

Most teams I work with add agent skills and MCP servers the way we once added npm packages: copied from a README, rarely read. A skill is a set of instructions your agent follows with your access. In February, Snyk scanned 3,984 public skills and found at least one security flaw in 36.82% of them, and malicious payloads in 76: https://snyk.io/blog/toxicskills-malicious-ai-agent-skills-clawhub/

Spellwarden checks them before your agents load them. The open-source scanner reads a skill, plugin or MCP server without running it, and flags prompt injection, tool poisoning, leaked secrets and unpinned packages. Each finding names the rule, the file and the line.

Two ways to try it:

pipx run spellwarden scan .

Live demo, no account needed: https://dash.spellwarden.ai/live-demo

Before the launch, I'd like to hear how you handle this today:

1. Do you read a skill or MCP server before you install it, or do you trust the source?

2. Do you pin versions, or take the latest?

3. What would a check need to show you before you trusted its result?

I'll read every answer. The checks you say are missing will shape the next rules.