CVE-2026-124028
mistral.rs before 0.9.5 contains a delimiter injection vulnerability in mistralrs-core that allows attackers to inject tool calls by placing delimiter strings …
Impact
mistral.rs before 0.9.5 contains a delimiter injection vulnerability in mistralrs-core that allows attackers to inject tool calls by placing delimiter strings in text the model quotes. Attackers controlling documents, web pages or tool outputs can embed tags like </tool_call> to make qwen-xml, gemma4 or reasoning parsers return extra calls to declared tools with attacker-chosen arguments.
An attacker can exploit this over the network, without logging in, but a user has to do something, such as open a file or a link. Specific conditions have to be met, which makes an attack harder. If it succeeds: low risk of data being read, low risk of data or systems being changed.
Affected products · EricLBuehler
| Product | Affected versions |
|---|---|
| Mistral.rs | before 0.9.5 |
Weakness classification
References
- NVD record
- CVE.org record
- GitHub advisory
- https://github.com/EricLBuehler/mistral.rs/commit/b43d3443e946752c0f662c2df42a2974410c9532
- https://github.com/EricLBuehler/mistral.rs
- https://github.com/EricLBuehler/mistral.rs/security/advisories/GHSA-34w8-j94f-8c5w
- https://github.com/EricLBuehler/mistral.rs/pull/2455
- https://github.com/EricLBuehler/mistral.rs/releases/tag/v0.9.5
- https://www.vulncheck.com/advisories/mistral-rs-before-0.9.5-tool-call-injection-via-mistralrs-core-output-parsers