MCP-Shield
Security and privacy
mcp server security scanner
SHOW HN2025
How HN took it
3 positive, 3 neutral or asking, 1 mixed, 7 critical, out of 14 comments about the project. Thread score -0.23.
Idea: 2 for, 3 againstExecution: 0 for, 4 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Useful space for security tooling
From 1 comment
Nice! This is a much-needed space for security tooling, and I appreciate that you've put some thought into the new attack vectors. I also like the combination of signature-based…
khafra on HN
Good that it combines pattern matching with LLM analysis
From 1 comment
Nice! This is a much-needed space for security tooling, and I appreciate that you've put some thought into the new attack vectors. I also like the combination of signature-based…
khafra on HN
Objections
Regex denylists are easy to bypass
From 1 comment
Sorry, but this will never work very well. The tool contains a bunch of "denylist regexes", i.e. ``` `user (should not|must not|cannot) see` ``` But these can easily be bypassed.…
NicolaiS on HN
Analysis step uses an LLM that is itself open to prompt injection
From 2 comments
So the analysis is done with another call to claude with instructions like "You are a cybersecurity expert..." basically another level of extreme indirection with unpredictable…
hollowturtle on HNSorry, but this will never work very well. The tool contains a bunch of "denylist regexes", i.e. ``` `user (should not|must not|cannot) see` ``` But these can easily be bypassed.…
NicolaiS on HN
Does not scan prompt injection in tool results
From 1 comment
Cool. If I'm not wrong you don't detect prompt injection done in the tool results? Any plans for that?
pcwelder on HN
Server can report different tools to scanner than to clients
From 1 comment
Neat, but what’s to stop a server from reporting one innocuous set of tools to MCP-Shield and then a different set of tools to the client?
paulgb on HN
Requests
Scan resources as well as tool definitions
From 1 comment
This is pretty cool. You should also attempt to scan resources if possible. Similar to the tool injection attack Invariant Labs discovered, I achieved the same result via…
mlenhard on HN
Integrate with kgateway
From 1 comment
Suggestion: Integrate with https://kgateway.dev/
stpedgwdgfhgdd on HN
Detect prompt injection in tool results
From 1 comment
Cool. If I'm not wrong you don't detect prompt injection done in the tool results? Any plans for that?
pcwelder on HN
Questions
Will it detect injection in non-English tool descriptions
From 1 comment
It seems that writing a tool in anything else than English will bypass most of this scanner
calyhre on HN
Alternatives named
mcp-scan by Invariant Labs
From 1 comment
Cool work! Thanks for citing our (InvariantLabs) blog posts! I really like the identify-as feature! We recently launched a similar tool ourselfs, called mcp-scan:…
marcfisc on HN
vet by SafeDep
From 1 comment
May be try out vet as well: https://github.com/safedep/vet vet is backed by a code analysis engine that performs malicious package (npm, pypi etc.) scanning. We recently extended…
abhisek on HN