io.github.cyanheads/met-museum-mcp-server

The only listed environment variable is MCP_LOG_LEVEL, which looks like a logging configuration rather than a sensitive secret; no API token, OAuth credential, or high-value account secret is indicated, so credential leakage or abuse risk appears low.

The tool is declared to connect to met-museum.caseyjhand.com, which is a clear remote network egress path. Based on the stated functionality, user queries may be sent to that endpoint to retrieve museum data; there is no evidence of unrelated third-party exfiltration or hidden endpoints, but this should still be treated as normal outbound data flow and monitored.

The system checks indicate that this MCP tool executes code or runs a local process, which is a normal property of this class of tool. The available material does not show it requesting system privileges beyond what would be expected for museum data retrieval, and there is no suspicious command-execution behavior described, so this is a caution rather than a clear high risk.

The description only refers to accessing MET Museum collections, metadata, provenance, and open-access images; there is no stated need to read local files, write to user data directories, or access additional local resources. Based on the available material, there is no sign of excessive data access.

Positive factors include: official registry listing, open-source code that can be audited, and updates within the last year. Caution points are the missing README, undeclared license, and low community adoption (0 stars), which reduce external scrutiny and maturity signals; however, the current evidence does not justify a high-risk rating.