A coding agent that asks “do you trust this repository?” is not much use if it has already run the repository’s commands. The Injection‘s week-45 roundup reports that VulnCheck published CVE-2026-122135 (CVSS 8.6) on November 4, 2026, for Mistral Vibe: the tool “ran commands from a repository’s own Git configuration at session start, before the trust prompt appeared.” Francisco Rosales of Manifold Security found three such flaws. All three are fixed in version 3.0.3.
Quick Take
- Product: Mistral Vibe, Mistral’s coding-agent tooling.
- Bug: Commands taken from a repo’s Git configuration executed at session start, ahead of the trust prompt. Three related flaws; CVE-2026-122135 is scored 8.6.
- Finder: Francisco Rosales, Manifold Security.
- Fix: Upgrade to Mistral Vibe 3.0.3 or later. Do not open untrusted repositories in older versions.
What happened
Trust prompts are a standard pattern in agentic IDEs and CLIs: the tool inspects a new repo, then asks the human before it will follow project-level hooks, configs or auto-run instructions. The Vibe bugs broke that order of operations. If Git configuration in the repo itself can run a command at session start, an attacker only needs the developer to point Vibe at a malicious or trojaned project. The prompt arrives too late.
This is the same family of failure as untrusted-repo bugs in other coding agents this season (Aider’s .aider.conf.yml issue is one example). The right fix is not a clearer dialog; it is to parse and apply repo-supplied configuration only after an explicit trust decision, and to ignore hook-like Git settings until then.
What you should do
- Update Mistral Vibe to 3.0.3 or later.
- Until you have updated, do not run Vibe against repositories you do not already trust, including fresh clones of unfamiliar GitHub projects and student or vendor sample repos.
- If an older Vibe session was started in an untrusted repo, treat that workstation as possibly exposed and review what the session ran.
What it means
Every major coding agent is being asked the same question this autumn: does “trust this workspace” actually gate execution, or is it a courtesy screen? A CVSS 8.6 on a pre-prompt execution path is a serious answer. Mistral shipped a fix in 3.0.3; teams should not assume other agents with a similar prompt are safe without checking the same order-of-operations bug.
About the analysis
Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated illustration. Read our AI Use Policy. Spotted an error? Request a correction.
This i10x report is based on The Injection’s week-45 summary of the VulnCheck advisory for CVE-2026-122135 (November 4, 2026). i10x did not test Mistral Vibe.
