Configure agent behavior
Agents can be configured using instruction files in your project that define how they should behave in your codebase, including coding conventions, architectural constraints, and common workflows.
Most agents rely on an AGENTS.md file for reusable guidance, while some agents use their own formats. For example, Claude Agent uses CLAUDE.md and the .claude directory. AI Assistant supports these instruction formats and passes them to the selected agent.
Agents automatically read these files and use them as guidance when generating code, executing commands, and making decisions. This helps keep project-specific instructions in one place and ensures more consistent results.
Instruction files vs Project rules
AI Assistant also provides a similar functionality called Project rules, which define how AI Assistant behaves in chat.
Instruction files (
AGENTS.md,CLAUDE.md) are used by the selected coding agent and travel with your repository.Project rules are configured in the IDE and apply only in AI Assistant chat mode.
Use instruction files to share guidance across tools and environments, and Project rules to customize behavior within AI Assistant.
How to use it
Create an instruction file supported by your agent (for example, AGENTS.md or CLAUDE.md) in the root of your repository and describe how agents should work with your codebase. The selected agent will automatically detect and use these instructions.
A typical file includes:
Project context – languages, frameworks, architecture.
Development rules – coding standards and constraints.
Repository conventions – directory structure and boundaries.
Common tasks – commands for building, testing, and running the project.
Restrictions – what the agent must not do.
Definition of done – conditions for completing a task.