PT-2026-53461 · Pypi · Langchain-Core

Publicado

2026-06-29

·

Atualizado

2026-06-29

CVSS v3.1

9.3

Crítica

VetorAV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N

Summary

A serialization injection vulnerability exists in LangChain's dumps() and dumpd() functions. The functions do not escape dictionaries with 'lc' keys when serializing free-form dictionaries. The 'lc' key is used internally by LangChain to mark serialized objects. When user-controlled data contains this key structure, it is treated as a legitimate LangChain object during deserialization rather than plain user data.

Attack surface

The core vulnerability was in dumps() and dumpd(): these functions failed to escape user-controlled dictionaries containing 'lc' keys. When this unescaped data was later deserialized via load() or loads(), the injected structures were treated as legitimate LangChain objects rather than plain user data.
This escaping bug enabled several attack vectors:
  1. Injection via user data: Malicious LangChain object structures could be injected through user-controlled fields like metadata, additional kwargs, or response metadata
  2. Class instantiation within trusted namespaces: Injected manifests could instantiate any Serializable subclass, but only within the pre-approved trusted namespaces (langchain core, langchain, langchain community). This includes classes with side effects in init (network calls, file operations, etc.). Note that namespace validation was already enforced before this patch, so arbitrary classes outside these trusted namespaces could not be instantiated.

Security hardening

This patch fixes the escaping bug in dumps() and dumpd() and introduces new restrictive defaults in load() and loads(): allowlist enforcement via allowed objects="core" (restricted to [serialization mappings](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain core/load/mapping.py)), secrets from env changed from True to False, and default Jinja2 template blocking via init validator. These are breaking changes for some use cases.

Who is affected?

Applications are vulnerable if they:
  1. Use astream events(version="v1") — The v1 implementation internally uses vulnerable serialization. Note: astream events(version="v2") is not vulnerable.
  2. Use Runnable.astream log() — This method internally uses vulnerable serialization for streaming outputs.
  3. Call dumps() or dumpd() on untrusted data, then deserialize with load() or loads() — Trusting your own serialization output makes you vulnerable if user-controlled data (e.g., from LLM responses, metadata fields, or user inputs) contains 'lc' key structures.
  4. Deserialize untrusted data with load() or loads() — Directly deserializing untrusted data that may contain injected 'lc' structures.
  5. Use RunnableWithMessageHistory — Internal serialization in message history handling.
  6. Use InMemoryVectorStore.load() to deserialize untrusted documents.
  7. Load untrusted generations from cache using langchain-community caches.
  8. Load untrusted manifests from the LangChain Hub via hub.pull.
  9. Use StringRunEvaluatorChain on untrusted runs.
  10. Use create lc store or create kv docstore with untrusted documents.
  11. Use MultiVectorRetriever with byte stores containing untrusted documents.
  12. Use LangSmithRunChatLoader with runs containing untrusted messages.
The most common attack vector is through LLM response fields like additional kwargs or response metadata, which can be controlled via prompt injection and then serialized/deserialized in streaming operations.

Impact

Attackers who control serialized data can extract environment variable secrets by injecting {"lc": 1, "type": "secret", "id": ["ENV VAR"]} to load environment variables during deserialization (when secrets from env=True, which was the old default). They can also instantiate classes with controlled parameters by injecting constructor structures to instantiate any class within trusted namespaces with attacker-controlled parameters, potentially triggering side effects such as network calls or file operations.
Key severity factors:
  • Affects the serialization path - applications trusting their own serialization output are vulnerable
  • Enables secret extraction when combined with secrets from env=True (the old default)
  • LLM responses in additional kwargs can be controlled via prompt injection

Exploit example

python
from langchain core.load import dumps, load
import os

# Attacker injects secret structure into user-controlled data
attacker dict = {
  "user data": {
    "lc": 1,
    "type": "secret",
    "id": ["OPENAI API KEY"]
  }
}

serialized = dumps(attacker dict) # Bug: does NOT escape the 'lc' key

os.environ["OPENAI API KEY"] = "sk-secret-key-12345"
deserialized = load(serialized, secrets from env=True)

print(deserialized["user data"]) # "sk-secret-key-12345" - SECRET LEAKED!

Security hardening changes (breaking changes)

This patch introduces three breaking changes to load() and loads():
  1. New allowed objects parameter (defaults to 'core'): Enforces allowlist of classes that can be deserialized. The 'all' option corresponds to the list of objects [specified in mappings.py](https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain core/load/mapping.py) while the 'core' option limits to objects within langchain core. We recommend that users explicitly specify which objects they want to allow for serialization/deserialization.
  2. secrets from env default changed from True to False: Disables automatic secret loading from environment
  3. New init validator parameter (defaults to default init validator): Blocks Jinja2 templates by default

Migration guide

No changes needed for most users

If you're deserializing standard LangChain types (messages, documents, prompts, trusted partner integrations like ChatOpenAI, ChatAnthropic, etc.), your code will work without changes:
python
 from langchain core.load import load

# Uses default allowlist from serialization mappings
obj = load(serialized data)

For custom classes

If you're deserializing custom classes not in the serialization mappings, add them to the allowlist:
python
from langchain core.load import load
from my package import MyCustomClass

# Specify the classes you need
obj = load(serialized data, allowed objects=[MyCustomClass])

For Jinja2 templates

Jinja2 templates are now blocked by default because they can execute arbitrary code. If you need Jinja2 templates, pass init validator=None:
python
from langchain core.load import load
from langchain core.prompts import PromptTemplate

obj = load(
  serialized data,
  allowed objects=[PromptTemplate],
  init validator=None
)
[!WARNING] Only disable init validator if you trust the serialized data. Jinja2 templates can execute arbitrary Python code.

For secrets from environment

secrets from env now defaults to False. If you need to load secrets from environment variables:
python
 from langchain core.load import load

obj = load(serialized data, secrets from env=True)

Credits

  • Dumps bug was reported by @yardenporat
  • Changes for security hardening due to findings from @0xn3va and @VladimirEliTokarev

Correção

Encontrou algum problema na descrição? Tem algo a acrescentar? Fique à vontade para nos escrever 👾
dbugs@ptsecurity.com

Identificadores relacionados

PYSEC-2026-373

Produtos afetados

Langchain-Core