Watching this talk about Agentic Search for Context Engineering by @helloiamleonie Watched half of this talk. Really well put together. It’s about how agents actually decide what to pull from files, databases, memory, and the web & why that choice often matters more than the model itself. You can get the slides here from her talk: https://github.com/iamleonie/workshop-agentic-search.git…

iamleonie/workshop-agentic-search

Source: https://github.com/iamleonie/workshop-agentic-search

Workshop: Agentic Search for Context Engineering

This workshop discusses different agentic search techniques for context engineering.

Learning outcomes: By the end of this workshop you will have

• experimented with a few different ways to search for context across different context sources
• learned how you can expand the capabilities of search tools with Agent Skills and additional CLIs
• gained an intuition on the trade-offs for different search tool

Set up

1. Create and activate a Python virtual environment

python3 -m venv .venv
source .venv/bin/activate

2. Install depencendies

This notebook uses langchain v1.2.12 and langchain-openai v1.1.11.

pip install -r requirements.txt 

3. Set up Elasticsearch

This workshop uses a local Elasticsearch instance as one of the out-of-context sources.

The easiest way to run Elasticsearch locally is using the start-local script, which sets up Elasticsearch (and optionally Kibana) using Docker with a simple one-line command:

curl -fsSL https://elastic.co/start-local | sh

This creates an elastic-start-local folder containing configuration files and startup scripts. To start Elasticsearch:

cd elastic-start-local
./start.sh

Expected output:

 ✔ Container es-local-dev          Healthy                                                                     
 ✔ Container kibana-local-dev      Healthy

4. Environment variables and API Keys

This workshop requires three categories of environment variables, which need to be added to the .env file (see .env.example)

• LLM API key: These notebooks use OpenAI models through LiteLLM. Alternatively, you can use any LLM of your choice that is capable of tool use.
• Jina API key: You can obtain a free Jina API key on from the official Jina page without registration.
• Elasticsearch credentials: During the set up of your local Elasticsearch instance, you will find the Password (username and url stay the same).

5. Data preparation

The workshop’s examples are based on the AI Engineer Europe Conference schedule, which is available in the data folder under session.json (Downloaded from: https://www.ai.engineer/europe/schedule).

This workshop discusses data store and filesystem as context sources. To prepare the Elasticsearch data store and the local filesystem, run the data preparation notebook.

Course Outline

Additional Resources

• The shell tool is not a silver bullet for context engineering
• Building effective database retrieval tools for context engineering
• Elasticsearch Quickstart
• Elastic Agent Skills
• jina-grep CLI