CVE-2026-34070 is a path traversal vulnerability in LangChain affecting langchain-core prior to version 1.2.22. Multiple functions in langchain_core.prompts.loading read files from filesystem paths embedded in deserialized prompt configuration dictionaries without validating against directory traversal sequences or absolute path injection. When an application passes user-influenced data to load_prompt() or load_prompt_from_config(), an attacker can supply a crafted prompt template or configuration that causes the framework to open unintended local files. File access is constrained by the implementation’s extension checks, including text templates and structured example files, but those checks do not prevent traversal to arbitrary locations on the host filesystem. The flaw enables unauthorized local file disclosure through the prompt-loading API.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No valid public exploits. Mallory filtered out 1 candidate as fakes, detection scripts, or README-only repos.
All candidate exploits were filtered out by Mallory's validation.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
16 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A langchain-core path traversal vulnerability in legacy load_prompt functions.
A directory traversal vulnerability in LangChain that allows arbitrary file access and may lead to command execution or exposure of sensitive information.
A path traversal vulnerability in LangChain's prompt-loading API that can allow arbitrary file access.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.