Imported from aimultiple-benchmark/langchain-bench-cur-2 (AGENTS.md). Install upstream with npx skills add aimultiple-benchmark/langchain-bench-cur-2. Copyright stays with the author.

Global Development Guidelines for LangChain Projects

Core Development Principles

1. Maintain Stable Public Interfaces ⚠️ CRITICAL

Always attempt to preserve function signatures, argument positions, and names for exported/public methods.

❌ Bad - Breaking Change:

def get_user(id, verbose=False):  # Changed from `user_id`
    pass

✅ Good - Stable Interface:

def get_user(user_id: str, verbose: bool = False) -> User:
    """Retrieve user by ID with optional verbose output."""
    pass

Before making ANY changes to public APIs:

• Check if the function/class is exported in __init__.py
• Look for existing usage patterns in tests and examples
• Use keyword-only arguments for new parameters: *, new_param: str = "default"
• Mark experimental features clearly with docstring warnings (using MkDocs Material admonitions, like !!! warning)

🧠 Ask yourself: "Would this change break someone's code if they used it last week?"

2. Code Quality Standards

All Python code MUST include type hints and return types.

❌ Bad:

def p(u, d):
    return [x for x in u if x not in d]

✅ Good:

def filter_unknown_users(users: list[str], known_users: set[str]) -> list[str]:
    """Filter out users that are not in the known users set.

    Args:
        users: List of user identifiers to filter.
        known_users: Set of known/valid user identifiers.

    Returns:
        List of users that are not in the known_users set.
    """
    return [user for user in users if user not in known_users]

Style Requirements:

• Use descriptive, self-explanatory variable names. Avoid overly short or cryptic identifiers.
• Attempt to break up complex functions (>20 lines) into smaller, focused functions where it makes sense
• Avoid unnecessary abstraction or premature optimization
• Follow existing patterns in the codebase you're modifying

3. Testing Requirements

Every new feature or bugfix MUST be covered by unit tests.

Test Organization:

• Unit tests: tests/unit_tests/ (no network calls allowed)
• Integration tests: tests/integration_tests/ (network calls permitted)
• Use pytest as the testing framework

Test Quality Checklist:

• Tests fail when your new logic is broken
• Happy path is covered
• Edge cases and error conditions are tested
• Use fixtures/mocks for external dependencies
• Tests are deterministic (no flaky tests)

Checklist questions:

• Does the test suite fail if your new logic is broken?
• Are all expected behaviors exercised (happy path, invalid input, etc)?
• Do tests use fixtures or mocks where needed?

def test_filter_unknown_users():
    """Test filtering unknown users from a list."""
    users = ["alice", "bob", "charlie"]
    known_users = {"alice", "bob"}

    result = filter_unknown_users(users, known_users)

    assert result == ["charlie"]
    assert len(result) == 1

4. Security and Risk Assessment

Security Checklist:

• No eval(), exec(), or pickle on user-controlled input
• Proper exception handling (no bare except:) and use a msg variable for error messages
• Remove unreachable/commented code before committing
• Race conditions or resource leaks (file handles, sockets, threads).
• Ensure proper resource cleanup (file handles, connections)

❌ Bad:

def load_config(path):
    with open(path) as f:
        return eval(f.read())  # ⚠️ Never eval config

✅ Good:

import json

def load_config(path: str) -> dict:
    with open(path) as f:
        return json.load(f)

5. Documentation Standards

Use Google-style docstrings with Args section for all public functions.

❌ Insufficient Documentation:

def send_email(to, msg):
    """Send an email to a recipient."""

✅ Complete Documentation:

def send_email(to: str, msg: str, *, priority: str = "normal") -> bool:
    """
    Send an email to a recipient with specified priority.

    Args:
        to: The email address of the recipient.
        msg: The message body to send.
        priority: Email priority level (`'low'`, `'normal'`, `'high'`).

    Returns:
        `True` if email was sent successfully, `False` otherwise.

    Raises:
        `InvalidEmailError`: If the email address format is invalid.
        `SMTPConnectionError`: If unable to connect to email server.
    """

Documentation Guidelines:

• Types go in function signatures, NOT in docstrings
  • If a default is present, DO NOT repeat it in the docstring unless there is post-processing or it is set conditionally.
• Focus on "why" rather than "what" in descriptions
• Document all parameters, return values, and exceptions
• Keep descriptions concise but clear
• Ensure American English spelling (e.g., "behavior", not "behaviour")

📌 Tip: Keep descriptions concise but clear. Only document return values if non-obvious.

6. Architectural Improvements

When you encounter code that could be improved, suggest better designs:

❌ Poor Design:

def process_data(data, db_conn, email_client, logger):
    # Function doing too many things
    validated = validate_data(data)
    result = db_conn.save(validated)
    email_client.send_notification(result)
    logger.log(f"Processed {len(data)} items")
    return result

✅ Better Design:

@dataclass
class ProcessingResult:
    """Result of data processing operation."""
    items_processed: int
    success: bool
    errors: List[str] = field(default_factory=list)

class DataProcessor:
    """Handles data validation, storage, and notification."""

    def __init__(self, db_conn: Database, email_client: EmailClient):
        self.db = db_conn
        self.email = email_client

    def process(self, data: List[dict]) -> ProcessingResult:
        """Process and store data with notifications."""
        validated = self._validate_data(data)
        result = self.db.save(validated)
        self._notify_completion(result)
        return result

Design Improvement Areas:

If there's a cleaner, more scalable, or simpler design, highlight it and suggest improvements that would:

• Reduce code duplication through shared utilities
• Make unit testing easier
• Improve separation of concerns (single responsibility)
• Make unit testing easier through dependency injection
• Add clarity without adding complexity
• Prefer dataclasses for structured data

Development Tools & Commands

Package Management

# Add package
uv add package-name

# Sync project dependencies
uv sync
uv lock

Testing

# Run unit tests (no network)
make test

# Don't run integration tests, as API keys must be set

# Run specific test file
uv run --group test pytest tests/unit_tests/test_specific.py

Code Quality

# Lint code
make lint

# Format code
make format

# Type checking
uv run --group lint mypy .

Dependency Management Patterns

Local Development Dependencies:

[tool.uv.sources]
langchain-core = { path = "../core", editable = true }
langchain-tests = { path = "../standard-tests", editable = true }

For tools, use the @tool decorator from langchain_core.tools:

from langchain_core.tools import tool

@tool
def search_database(query: str) -> str:
    """Search the database for relevant information.

    Args:
        query: The search query string.
    """
    # Implementation here
    return results

Commit Standards

Use Conventional Commits format for PR titles:

• feat(core): add multi-tenant support
• fix(cli): resolve flag parsing error
• docs: update API usage examples
• docs(openai): update API usage examples

Framework-Specific Guidelines

• Follow the existing patterns in langchain-core for base abstractions
• Use langchain_core.callbacks for execution tracking
• Implement proper streaming support where applicable
• Avoid deprecated components like legacy LLMChain

Partner Integrations

• Follow the established patterns in existing partner libraries
• Implement standard interfaces (BaseChatModel, BaseEmbeddings, etc.)
• Include comprehensive integration tests
• Document API key requirements and authentication

Quick Reference Checklist

Before submitting code changes:

• Breaking Changes: Verified no public API changes
• Type Hints: All functions have complete type annotations
• Tests: New functionality is fully tested
• Security: No dangerous patterns (eval, silent failures, etc.)
• Documentation: Google-style docstrings for public functions
• Code Quality: make lint and make format pass
• Architecture: Suggested improvements where applicable
• Commit Message: Follows Conventional Commits format