MCP Observatory
3m ago Tracked 26,716 Last 3m ago 24h 1,578
npm JavaScript analyzed 0.2.2 unconfirmed MCP

@21st-dev/magic

v0.2.2
npm

Magic MCP is now the 21st MCP. This package is a compatibility proxy kept for old configs - new setups should use `npx @21st-dev/cli@latest init`. Setup: https://21st.dev/mcp

maintainer
bunasq
license
ISC
first seen
2026-05-22
last seen
2026-08-02
releases · 30d
3
short id
risk 31/100 · heuristic grade
B low
  • capability exposure inferred + 24
  • inherited (deps) attested + 15
  • trust mitigators mixed − 8

attested inferred mixed

The A–E grade is our heuristic synthesis — a "review this" prompt, not a verdict. Each factor is tagged by what backs it: attested (a verifiable record), reported (a third party's claim), or inferred (our own heuristic, e.g. permissions). See methodology.

graded 39s ago · see ecosystem CVEs →

risk trajectory 4 movements
  • C · 35 → B · 31
  • C · 38 → C · 35
  • C · 45 → C · 38
  • C · 38 → C · 45
capability exposure grade factor +24
Inferred surface — each links to servers holding it:
vulnerabilities 0 CVEs

No known CVEs for this server.

inherited (deps) · grade factor +15

inherited via dependencies 3
HIGH
@modelcontextprotocol/sdk has cross-client data leak via shared server/transport instance reuse via @modelcontextprotocol/sdk · 1 hop CVE-2026-25536 fixed in 1.26.0
HIGH
Anthropic's MCP TypeScript SDK has a ReDoS vulnerability via @modelcontextprotocol/sdk · 1 hop CVE-2026-0621 fixed in 1.25.2
HIGH
Model Context Protocol (MCP) TypeScript SDK does not enable DNS rebinding protection by default via @modelcontextprotocol/sdk · 1 hop CVE-2025-66414 fixed in 1.24.0
tool safety all quiet

No tool-safety findings — heuristic detectors run on the compute-risk cadence; a finding appears when a tool trips a rule.

embed badge readme-ready
live risk-grade badge preview [![MCP Observatory risk grade](https://mcpobservatory.com/servers/npm:@21st-dev/magic/badge.svg)](https://mcpobservatory.com/servers/npm:@21st-dev/magic/security)

Heuristic, inferred signals — false positives (legitimately powerful tools, forks, language ports) are expected. Treat each as "review this", not a verdict. See the ecosystem-wide picture on the security hub, or the fleet security of bunasq.