Meta thought about ADK-Go and ADK-Docs

[verdverm]

ADK is a framework that started not in Go

• the docs do not have as many Go examples
• the paragraphs of text use python_casing instead of TitleCasing when referring to core concept interfaces when describing how they work

Now as someone who is building agentic stuff with ADK-Go with the help of a coding agent (also built with adk)... I want to provide some context on how this core component of my arch works. Something like: https://google.github.io/adk-docs/runtime/

The language difference is extra work for the LLM or Users on a case-by-case basis.

1. I believe there is already discussion and/or work towards the Go docs gap
2. What can we do to make it easier for developers (users of ADK) to get helpful context based on their language and coding agent setup?

[verdverm]

Weird how how writing docs is changing, I'm altering the markdown at the runtime page to

1. rename inline names to the Go ones, removing mentions to other languages
2. making sure "references" are consistent

first we shape our tools, then our tools shape us

from the runtime page, before"

### Runner's Role (Orchestrator)

The `Runner` acts as the central coordinator for a single user invocation. Its responsibilities in the loop are:

...

3. **Receive & Process:** Waits for the agent logic to `yield` an `Event`. Upon receiving an event, the Runner **promptly processes** it. This involves:

...

`Runner` to complete step 3 (processing and committing)

from the runtime page, after:

3. **Processing & Committing:** Waits for the agent logic to `yield` an `Event`. Upon receiving an event, the Runner **promptly processes** it. This involves:

...

`Runner` to complete step 3 (Processing & Committing)

this makes sure the reference in the paragraph aligns with the bullet point "label", important for cajoling the attention mechanisms

[dpasiukevich]

cc @ivanmkc @tpryan

[tpryan]

Also like to include @joefernandez

[verdverm]

For fun, I had my coding agent built on ADK, using gemini-2.5-flash, generate a bunch of AGENTS.md files as a first step towards assembling Go specific agent instruction files for context. Primarily focused on the ./internal dir, but also generated a root AGENTS.md after looking at all the other dirs and the internal agent files.

I still have to go through them all to see if they are accurate, is this my new life with agents... prompting and double checking agent work? 🫠

verdverm@a44dedf