AI Coding Agent & Editor Setup¶
How to configure AI coding agents and code editors to generate accurate adk-fluent and Google ADK code.
AI coding agents work best when they have access to up-to-date API documentation and project conventions. The pages below show you how to wire that context into your tool of choice — using rules files, MCP servers, or both.
Pick your tool¶
CLI-based AI coding agent from Anthropic. Uses CLAUDE.md for project rules and supports MCP servers.
AI-first code editor. Uses .cursor/rules/ for project context and supports MCP servers.
GitHub Copilot in VS Code. Uses .github/instructions/ for project context and supports MCP via .vscode/mcp.json.
AI code editor by Codeium. Uses .windsurfrules for project context and supports MCP servers.
VS Code extension. Uses .clinerules/ for project context and supports MCP servers.
High-performance editor. Uses llms.txt via #fetch and supports MCP context servers.
14 portable skills for developing, testing, debugging, deploying, evaluating, and reviewing adk-fluent projects. Works with any Agent Skills-compatible platform.
How it works¶
Each setup page covers two complementary approaches:
- Rules files
A project-level file (e.g.
CLAUDE.md,.cursorrules) that tells the AI agent about adk-fluent’s API patterns, conventions, and best practices. This is the most impactful step — it teaches the agent how to write idiomatic adk-fluent code.- MCP servers
A Model Context Protocol server that gives the AI agent live access to adk-fluent’s full documentation. This lets the agent look up exact method signatures, builder options, and cookbook recipes on demand.
Both approaches can be used independently, but combining them gives the best results.
All rules files (CLAUDE.md, .cursor/rules/adk-fluent.mdc, .windsurfrules, etc.) are auto-generated from the same source of truth by scripts/llms_generator.py. They update automatically when the API changes, so they never go stale. Run just llms to regenerate them locally, or let CI handle it.
Connection to the three pathways¶
AI coding agents benefit from understanding which development pathway you’re using:
Pathway |
What the AI agent needs to know |
Key rules/skills |
|---|---|---|
Pipeline |
Expression operators, namespace modules, builder methods |
Rules files (auto-generated), |
Skills |
SKILL.md format, topology expressions, |
|
Harness |
H namespace, 5-layer architecture, safety patterns |
|
The rules files cover all three pathways. The Agent Skills provide deeper procedural guidance for each.