Skip to main content

QuickJS integration package for Deep Agents

Project description

langchain-quickjs

langchain-quickjs provides a QuickJS-backed REPL middleware for Deep Agents. It adds a repl tool that can evaluate small JavaScript snippets for computation, control flow, JSON manipulation, and calls to exposed Python foreign functions.

Basic usage

from deepagents import create_deep_agent
from langchain_quickjs import QuickJSMiddleware


def normalize_name(name: str) -> str:
    return name.strip().lower()


agent = create_deep_agent(
    model="openai:gpt-4.1",
    tools=[],
    middleware=[
        QuickJSMiddleware(
            ptc=[normalize_name],
            add_ptc_docs=True,
        )
    ],
)

With this middleware installed, the agent receives a repl tool that runs each JavaScript evaluation in a fresh QuickJS context. If you expose Python callables through ptc, they are available inside the REPL as foreign functions.

REPL behavior

  • The REPL is stateless. Each call starts from a fresh QuickJS context, so variables, functions, and other values defined in one repl call are not available in the next one.
  • Execution uses QuickJS, so JavaScript support is limited to what QuickJS provides. It is good for small computations, control flow, JSON manipulation, and calling exposed foreign functions, but it is not a browser or Node.js runtime and does not provide their APIs.
  • Foreign functions support passing primitive values between JavaScript and Python, including int, float, bool, str, and None. Lists and dictionaries returned from Python are also supported and are bridged back into JavaScript arrays and objects.
  • Async foreign functions are supported. Because QuickJS callbacks are synchronous, awaitables are delegated to a dedicated daemon-thread event loop and their resolved results are returned back into the REPL call.

Current limitations

  • Does not work with HIL in the REPL.
  • Does not support ToolRuntime yet.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

langchain_quickjs-0.0.1.tar.gz (136.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

langchain_quickjs-0.0.1-py3-none-any.whl (12.7 kB view details)

Uploaded Python 3

File details

Details for the file langchain_quickjs-0.0.1.tar.gz.

File metadata

  • Download URL: langchain_quickjs-0.0.1.tar.gz
  • Upload date:
  • Size: 136.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for langchain_quickjs-0.0.1.tar.gz
Algorithm Hash digest
SHA256 9a75fadda31eae075ba50d49c07c4a6c8244d9d3eba00998522ad52055f152ca
MD5 b93da24e65942258665a4d5fff56bd4e
BLAKE2b-256 5f0e6cf8c5a487ce76f6c8ea199cf6622bbbadf5f937d8d8d628efc8c7e9d4a7

See more details on using hashes here.

File details

Details for the file langchain_quickjs-0.0.1-py3-none-any.whl.

File metadata

File hashes

Hashes for langchain_quickjs-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 4d1c5caec4de370d963e6d0df898ce5f6f950b2b75cd841753b079ccb1d3e77b
MD5 8c4f3b92814831aae22f456c115ce31b
BLAKE2b-256 2472925cbad2460e07b00a9e4fa2be6c2e2fec2067feb2e2f906060e0be41548

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page