io.github.dl-eigenart/agentshield-mcp

MCPcommunity
v0.1.3io.github.dl-eigenartUnknownUpdated 2mo agonpmGitHub

Detect prompt injection, jailbreak, and social-engineering attacks in LLM agents.

Official MCP (Model Context Protocol) server for AgentShield — the runtime gateway and real-time classifier that detects prompt-injection, jailbreak, and social-engineering attempts in text while your agent is running, not in an offline audit pass. Works with any MCP-compatible client: Claude Desktop, Cursor, Cline, Zed, Continue, and custom agents. Single-shot per request, p50 ~2.4 ms — designed…

Automatically indexed from public sources. Not yet verified by the developer on Forge.Claim this listing →
2mo agoLast update
Package
Authorio.github.dl-eigenart
LicenseUnknown
Version0.1.3
Sourcemcp-registry
npm@eigenart/agentshield-mcp
GitHubdl-eigenart/agentshield-platform
Trust Status
B
60/100Good
✓Listed in Forge index+10/10
—Publisher identity verified+0/25
→ Publisher: run `forge publish` from the package repo to claim ownership
—Ed25519 publish signature+0/10
→ Included automatically when the publisher runs `forge publish`
—Domain verification+0/5
→ Publisher: host /.well-known/forge.json on the package homepage with { "publisher": "<github-login>" }
✓CVE scan · clean+30/30
✓Static analysis · clean+20/20
—npm provenance (Sigstore)+0/5
→ Publish from GitHub Actions with the --provenance flag
Paste into Claude Code, Cursor, or any AI assistant to fix all gaps
StatusCommunity-indexed
PublisherUnverified
SignatureUnsigned
Domain—
Provenance—
Dependencies✓ 60 resolved+ · none vulnerable
Tool surface1 tool · none privileged
Security scan✓ Cleanv0.1.6 · 20d ago
EvalsNone
IndexedJun 13, 2026

Verification confirms publisher identity (repo ownership), not code safety. The security scan covers known CVEs and suspicious install scripts — it cannot prove the absence of malicious code.

About

Official MCP (Model Context Protocol) server for AgentShield — the runtime gateway and real-time classifier that detects prompt-injection, jailbreak, and social-engineering attempts in text while your agent is running, not in an offline audit pass. Works with any MCP-compatible client: Claude Desktop, Cursor, Cline, Zed, Continue, and custom agents. Single-shot per request, p50 ~2.4 ms — designed to sit in the agent's hot path on every untrusted input. Exposes one tool to the agent: . Call it…

Keywords
mcp