Self: https://github.com/palladius/ai-friendly-agents/ (public)
This repo provide easy ready-to-use agents which work with the Python ADK library.
More on ADK:
Currently some agents are broken. I've added a dangerous change.
However, a stable version is available here: https://github.com/palladius/ai-friendly-agents/tree/v0.9/
A bunch of friendly agents ready to use with ADK, MCP, A2A et al.
Dir structure:
adk/prod/: agents built with ADK. Production ready, ready to use.adk/dev/: agents built with ADK. More to learn how to do things. Could be cut and paste of public docs.mcp-agents/prod: agents built with MCP. Production ready, ready to use.mcp-agents/dev: agents built with MCP. More to learn how to do things.rag/: samples to instruct AIs with local code.
Better samples can be found here: https://github.com/google/adk-samples
To test MCP agents:
- Nice locally running agent:
npx @modelcontextprotocol/inspector - Use the
pip install mcp[cli]. Never tried yet. Example.
First, install ADK for python: pip install google-adk
# 1. Clone and adapt env file
git clone https://github.com/palladius/ai-friendly-agents/
cp .env.dist .env
# edit away .env with your info
# 2. Install virtualenvenv
python -m venv .venv
source .venv/bin/activate
# 3.
cd adk/prod/
pip install -r requirements.txt