Know which AI tools you can trust — before you plug them in.
TrustBoard is the discovery layer for MCP servers and agent skills: search the whole ecosystem, read transparent trust reports, and show a verified badge on your README.
Weather MCP (example)
example.dev/weather-mcp · v2.4.0
The MCP ecosystem is exploding. Trust hasn’t kept up.
Thousands of servers can read your files, call APIs and run commands on your behalf. These charts are computed from the public registry — real data, refreshed with each snapshot.
Release activity by month
Servers grouped by the month their current version shipped
How servers are delivered
Primary transport / package type
What servers do
Keyword-based classification of names and descriptions (a server can match several categories)
Five checks. One transparent score.
Every report is built from public, verifiable evidence. No pay-to-rank: Featured placements never change a score.
License
OSI license detected, compatible with commercial use, no missing LICENSE file.
Maintenance
Release cadence, recent commits, open-issue response time.
Permissions
Tools, scopes and secrets the server asks for — file system, network, shell.
Known issues
Matches against public advisories and dependency CVEs.
Source verification
Published package matches the public repository and namespace owner.
Everything you need to decide — on one page.
Here’s what a finished report looks like. This one uses a fictional server and example data; real projects show “Pending review” until their scan is complete.
- Score breakdown across five weighted checks
- Exact permissions and secrets the server requests
- Advisory and dependency vulnerability matching
- Proof that the package matches its source repo
Weather MCP (example)
example.dev/weather-mcp
Show users your server is safe to install.
One line of Markdown. The badge updates automatically as your report changes.
weather-mcp
An MCP server that returns forecasts and alerts. Works with any MCP client.
[](https://trustboard.77705457.workers.dev/server?name=YOUR/SERVER)
# HTML
<img src="https://trustboard.77705457.workers.dev/badge.svg?name=YOUR/SERVER" alt="TrustBoard"/>
Free to list. Pay only to be seen.
Paid placements are clearly labeled and never affect trust scores.
Listing
For every MCP server and skill author.
- Catalog listing & search
- Public trust report
- Live README badge
Featured
Get discovered by developers searching right now.
- Homepage spotlight slot
- Pinned in keyword results
- Labeled “Sponsored” — scores untouched
- Click & impression analytics
Teams
Approve which MCP servers your company may use.
- Private allow-list
- Change alerts on approved servers
- Exportable PDF reports
Questions, answered.
What is an MCP server?
The Model Context Protocol lets AI assistants such as coding agents and chat apps connect to external tools and data. An MCP server exposes those tools — which is also why it deserves scrutiny before you install it.
Why do listings say “Pending review”?
We never show a number we can’t back with evidence. Each server keeps the Pending review label until its five checks have actually run.
Where does the catalog data come from?
Listings are pulled live from the official MCP Registry (registry.modelcontextprotocol.io). Authors can also submit servers directly.
Can I pay to get a better score?
No. Featured placements buy visibility only, are labeled as sponsored, and are kept completely separate from scoring.
Is a trust report a security audit?
No. Reports are automated, evidence-based signals to help you decide faster. For high-risk use, still review the code yourself.
Ship an MCP server? Get it in front of the right users.
Free listing, a live badge, and a trust report your users can actually read.