humanlayer/examples/langchain

LangChain Examples

Two ways to get started:

1. If you're comfortable with virtualenvs
2. If you prefer docker

If you're comfortable with virtualenvs

Activate a new virtualenv however you prefer, for example:

python3 -m venv venv
source venv/bin/activate

install requirements

pip install -r requirements.txt

python 01-math_example.py

If you prefer docker

docker compose run examples 01-math_example.py

All Examples

• 00-math_no_approvals.py - A simple math example that does not use humanlayer
• 00a-math_example_cli.py - A simple math example that uses humanlayer in CLI (offline) mode to enforce approvals
• 01-math_example.py - A simple math example that uses HumanLayer to gate access to the multiply function
• 02-math_with_eval.py - A math example that gates access to the run_python_code(code: str) -> str function
• 03-human_as_tool.py - A tou example of using HumanLayer's human_as_tool() function to create a tool that the angent can use to contact a human for input
• 04-human_as_tool_onboarding.py - A more in-depth example of an agent that uses a human_as_tool channel to get feedback on emails that are to be sent to new customers
• 04-human_as_tool_linkedin.py - Human-as-tool for a linkedin inbox manager (also uses require_approval)
• 04-human_as_tool_linkedin_frustration.py - prompting an agent to reach when a user sounds frustrated

• 05-approvals_and_humans_composite.py - Wrapping a "contact a human call" with an approval step routed to another human

• 06-custom_bot_token.py - Example of how to override the default slack bot token for specific ContactChannels