Playbooks · Coding tasks
A critical flaw in gemini-mcp-tool, fixed in 1.1.6
Written November 8, 2026
gemini-mcp-tool had a critical @file exfiltration and Windows command-injection flaw, fixed in 1.1.6. Who is affected, and the one-line check.
What the flaw is
GitHub's advisory GHSA-4h5r-5jm8-jxjm (CVE-2026-0755) records two flaws in gemini-mcp-tool, an MCP server that wraps the Gemini CLI. Untrusted prompt input could reach the Gemini CLI @file parser, allowing read and exfiltration of arbitrary local files — the advisory names @/etc/passwd, @~/.ssh/id_rsa, and @../../secret as examples. On Windows, unquoted cmd.exe metacharacters could break out into OS command injection. The Zero Day Initiative, which reported the command-injection flaw as ZDI-26-021, describes it as a flaw in the execAsync method: a user-supplied string was used to execute a system call without proper validation, and an attacker could leverage it to execute code in the context of the service account, with no authentication required.
The mechanism matters because of where the input comes from. Prompt input is untrusted by definition — it can arrive from a file the agent reads, a page it fetches, or text a user pastes in — and this flaw let that input reach a file parser and a shell. The @file side is the quieter of the two: a file read needs no command to run, only a reference the parser follows.
Who is affected, and the fix
- Affected versions: 1.1.2 and later, below 1.1.6.
- Patched version: 1.1.6. Upgrading to 1.1.6 or later closes the flaw.
- The fix in 1.1.6: the broken
shell:falsedouble-quote wrapping was removed;assertSafeFileReferences()was added to contain@filereferences to the working directory; Windowscmd.exeargument quoting was hardened. - Dates: the advisory was published May 30, 2026, and reviewed June 18, 2026.
The one-line check
Check the installed version, then upgrade if it is in the affected range:
npm list gemini-mcp-tool
npm install -g gemini-mcp-tool@latest
If the tool is installed as a dependency of something else rather than globally, check the version in that project's package.json or lockfile and upgrade there instead — a global upgrade does not fix a pinned local copy. If the version is below 1.1.2, it predates the affected range; if it is 1.1.6 or later, it carries the fix.
What not to assume
- The advisory does not report exploitation in the wild. Nothing in it says the flaw was used against anyone; the risk is the exposure, not an observed attack.
- Scoring is incomplete. NVD lists the Zero Day Initiative's CVSS vector (high impact on confidentiality, integrity, and availability) and notes that no NIST assessment has been provided yet. The severity rests on the advisory and the vector, not on a published score.
- The fix is version-scoped, not behaviour-scoped. Upgrading closes these two flaws. It does not make prompt input trustworthy in general — any tool that passes untrusted text to a file parser or a shell carries the same shape of risk.
Limits
- This page follows the GitHub advisory and the NVD record. No exploit was reproduced, and none should be: the check is the version number, not the flaw.
- The advisory covers gemini-mcp-tool only. Other MCP servers wrapping CLIs may share the pattern, but this page makes no claim about them.
Related playbooks
- AWS MCP server let deny lists silently fail
- Patch your MCP reference servers now
- Agenta 0.128.13: tool budgets extend the watchdog
- Manage Airbyte pipelines through its MCP server
Sources
- gemini-mcp-tool vulnerable to OS command injection and @file exfiltration via prompt quoting (CVE-2026-0755) — GitHub Advisory Database
- CVE-2026-0755 — NVD
- ZDI-26-021: gemini-mcp-tool execAsync Command Injection — Zero Day Initiative
Back to: Coding playbooks hub · Also: MCPhantom: agents install dead MCP servers · A malicious MCP server probed cloud credentials · 140 silent MCP contract changes, and how to pin them