Dependencies

This document explains StuntDouble’s dependencies and why each is required.

Core Dependencies

These are installed automatically with StuntDouble:

[project]
dependencies = [
    "langchain-core>=1.2.5",
    "pydantic>=2.0.0",
    "langgraph>=1.0.0",
]

langchain-core

Version: >=1.2.5

Why it’s needed:

Component	Usage
`StructuredTool`	Converting mirrored tools to LangChain format
`RunnableConfig`	Per-invocation configuration passing
Type definitions	Tool protocols and interfaces

Used by:

stuntdouble/langgraph/wrapper.py — create_mockable_tool_wrapper, mockable_tool_wrapper
stuntdouble/mirroring/ — LangChain tool conversion

pydantic

Version: >=2.0.0

Why it’s needed:

Component	Usage
`BaseModel`	Schema validation for tools
`create_model`	Dynamic schema generation from MCP tools
Field validation	Input/output type checking
JSON schema	Converting between formats

Used by:

stuntdouble/langgraph/wrapper.py — Dynamic schema conversion
stuntdouble/types.py — Type models (ScenarioMetadata, MockRegistration, etc.)
stuntdouble/langgraph/validation.py — Signature validation

langgraph

Version: >=1.0.0

Why it’s needed:

Component	Usage
`StateGraph`	Building mockable agent graphs
`ToolNode`	Native node with `awrap_tool_call` hook
`ToolCallRequest`	Tool call data extraction
Message types	State management compatibility

Used by:

stuntdouble/langgraph/wrapper.py — mockable_tool_wrapper, create_mockable_tool_wrapper

Development Dependencies

These are only needed for development and testing:

[dependency-groups]
dev = [
    "black >=25.9.0",
    "isort >=7.0.0,<8.0.0",
    "mypy >=1.18.1,<2.0.0",
    "pytest-cov>=7.0.0",
]
test = [
    "pytest >=8.4.0",
    "pytest-cov >=7.0.0,<8.0.0",
    "pytest-xdist >=3.8.0,<4.0.0",
    "tox >=4.26.0,<5.0.0",
    "tox-uv >=1.28.0",
]

Package	Purpose
`black`	Code formatting
`isort`	Import sorting
`mypy`	Static type checking
`pytest`	Test framework
`pytest-cov`	Coverage reporting
`pytest-xdist`	Parallel test execution
`tox`	Multi-environment testing
`tox-uv`	uv integration for tox

Optional Dependencies

These are NOT included but may be needed for specific features:

For LLM-powered mock generation

pip install langchain-openai
# or
pip install foundationsai

When needed: Using ToolMirror.with_llm() for realistic mock data generation.

For MCP mirroring support

pip install "stuntdouble[mcp]"

When needed: Using live MCP discovery and mirroring. The optional mcp extra currently provides the additional client dependency used by the StuntDouble MCP client.

Version Compatibility

Python Versions

StuntDouble supports:

Python 3.11
Python 3.12
Python 3.13
Python 3.14

LangChain Ecosystem

StuntDouble is designed around the LangChain Core / LangGraph 1.x series:

Package	Compatible Versions
`langchain-core`	1.x
`langgraph`	1.x

Note: Optional integrations such as langchain-openai are not pinned in StuntDouble’s package metadata. If you use them, align them with your project’s existing LangChain stack.

If you encounter dependency conflicts, ensure your project’s LangChain packages are aligned:

# Check versions
pip show langchain-core langgraph langchain-openai

# Upgrade to compatible versions
pip install --upgrade langchain-core>=1.2.5 langgraph>=1.0.0

Dependency Graph

┌─────────────────────────────────────────────────────────────────────────────┐
│                        StuntDouble Dependencies                             │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│                          stuntdouble                                        │
│                               │                                             │
│         ┌─────────┬───────────┬─────────┐                                  │
│         │         │           │         │                                  │
│         ▼         ▼           ▼         ▼                                  │
│   langchain-core  pydantic  langgraph  (optional)                          │
│         │           │           │         │                                │
│         │           │           │         ▼                                │
│         │           │                     │     langchain-openai           │
│         │           │                     │           │                    │
│         └───────────┴─────────────────────┴───────────┘                    │
│                               │                                             │
│                               ▼                                             │
│                   Common transitive deps:                                   │
│                   • typing-extensions                                       │
│                   • httpx/httpcore                                          │
│                   • tenacity                                                │
│                                                                             │
└─────────────────────────────────────────────────────────────────────────────┘

Minimal Installation

For the absolute minimal installation (core mocking only):

# These are the only truly required imports for basic mocking:
from stuntdouble.matching import InputMatcher
from stuntdouble.resolving import ValueResolver

The core matching and resolution logic has no external dependencies beyond the standard library. However, for practical use with LangChain/LangGraph agents, the full dependency set is recommended.

Troubleshooting

Dependency Conflicts

If you see errors like:

langchain-core 1.x requires ..., but you have ...

Solution:

# Pin versions explicitly
pip install langchain-core==1.2.5 langgraph==1.0.0
pip install stuntdouble

Missing Optional Dependencies

If you see:

ImportError: langchain-openai is not installed

Solution:

pip install langchain-openai