UTC
Medium

Safari MCP: JavaScript injection through single-quoted tool arguments (backslash not escaped)

GHSA-4rj4-rcrq-gr92 · published 2026-11-19

Severity
Medium
CVSS score
6.2
Chance of exploitation (EPSS)
No CVE
Fixed in
2.22.15
What it is

Sixteen places in safari.js placed a tool argument inside a single-quoted JavaScript string literal and escaped the single quote without first escaping backslashes. An argument containing a backslash followed by a quote therefore closed the literal, and the rest of the argument ran as JavaScript in the page. Seven of the affected arguments belong to read-only tools that are exempt from the tab-ownership check: filter in safari_network_details and safari_extract_links, level in safari_console_filter, selector in safari_snapshot, safari_accessibility_snapshot and safari_extract_tables, and dbName / storeName in safari_get_indexed_db. …

What to do

Upgrade safari-mcp to 2.22.15 or later. If you start it from an MCP config with npx or uvx, pin that version so an old cached copy is not used.

Fixed in 2.22.15: every affected site uses the shared escJsSingleQuote, which escapes the backslash before the quote, and a test fails the build if a quote-only escape returns.

If you cannot upgrade yet

Upgrade to 2.22.15. There is no configuration workaround: excluding safari_evaluate and safari_run_script from a client's tool list did not prevent the injection.

Affected
PackageRegistryAffectedFixed in
safari-mcpnpm< 2.22.152.22.15

The impact, patch and workaround text comes from the advisory, shortened (GitHub Advisory Database, CC BY 4.0). Proof-of-concept and reproduction sections and code are left out; the original advisory has them.

Sources: GitHub Advisory Database, reviewed and malware advisories (CC BY 4.0, https://github.com/advisories); advisories published by Cursor (https://github.com/cursor/cursor/security/advisories); advisories published by OpenClaw (https://github.com/openclaw/openclaw/security/advisories); advisories published by vLLM (https://github.com/vllm-project/vllm/security/advisories); advisories published by Open WebUI (https://github.com/open-webui/open-webui/security/advisories); advisories published by n8n (https://github.com/n8n-io/n8n/security/advisories); advisories published by Flowise (https://github.com/FlowiseAI/Flowise/security/advisories); advisories published by Langflow (https://github.com/langflow-ai/langflow/security/advisories); advisories published by the MCP Python SDK project (https://github.com/modelcontextprotocol/python-sdk/security/advisories); advisories published by MLflow (https://github.com/mlflow/mlflow/security/advisories); CISA Known Exploited Vulnerabilities catalog (CC0). Data as of 2026-11-22. ShieldMap describes how each problem works and how to fix it; it does not publish exploit code. Report a correction to [email protected].