How sync-agent-rules.sh Regenerates Platform-Specific Instruction Files for AI Agents

The sync-agent-rules.sh script maintains consistency across AI agent configurations by reading AGENTS.md, resolving file imports, and writing standardized instruction files to platform-specific locations.

In the JCodesMore/ai-website-cloner-template repository, managing AI agent instructions across multiple development platforms is automated through a single source of truth. The sync-agent-rules.sh utility ensures that changes to central agent rules propagate instantly to GitHub Copilot, Cline, Continue, and Amazon Q Developer without manual file duplication.

Locating the Source Configuration

The script determines the repository root and defines the canonical configuration path:

SOURCE=$REPO_ROOT/AGENTS.md

Safety checks (lines 23-32) abort execution immediately if AGENTS.md is missing, preventing incomplete regeneration of instruction files.

Resolving File Imports

The resolve_imports function (lines 33-50) handles Claude-style file includes. When AGENTS.md contains lines beginning with @, the script treats these as import directives:

• @path/to/file.md is replaced with the actual file contents
• The fully expanded document is stored in RESOLVED_CONTENT

This allows you to modularize agent rules across multiple files while maintaining a single compiled output.

Generating the Auto-Generated Header

Before writing outputs, the script constructs a standard header (lines 55-56) that marks files as derived from AGENTS.md. This prevents developers from editing generated files directly, ensuring all changes flow through the source document.

Writing Platform-Specific Files

The write_file helper function (lines 58-65) creates target directories recursively using mkdir -p if needed, then writes the header plus resolved content to each destination.

GitHub Copilot Chat

Outputs to .github/copilot-instructions.md as a direct copy of RESOLVED_CONTENT.

Cline / Roo Code

Receives the rules at .clinerules with identical content to the source document.

Continue

Requires special formatting at .continue/rules/project.md. The script prepends YAML front-matter (lines 75-81) before the resolved content to match Continue's configuration schema:

# Auto-generated from AGENTS.md

name: Project Rules
description: Generated project-specific instructions

Amazon Q Developer

Stores instructions at .amazonq/rules/project.md as a direct copy of RESOLVED_CONTENT.

Running the Regeneration Script

Execute the utility from the repository root:

bash scripts/sync-agent-rules.sh

The script outputs confirmation for each generated file:

Syncing agent rules from AGENTS.md...
  ✓ .github/copilot-instructions.md
  ✓ .clinerules
  ✓ .continue/rules/project.md
  ✓ .amazonq/rules/project.md

Done. Generated files are committed to the repo but sourced from AGENTS.md.
Edit AGENTS.md, then re-run this script to update all agent configs.

Summary

• Single source of truth: All agent instructions originate from AGENTS.md in the repository root.
• Import resolution: The resolve_imports function expands @file references to create a complete rule set.
• Multi-platform support: The script generates configuration files for GitHub Copilot, Cline, Continue, and Amazon Q Developer.
• Safety checks: Built-in validation ensures AGENTS.md exists before processing begins.
• Automation: Running bash scripts/sync-agent-rules.sh updates all platform-specific files simultaneously.

Frequently Asked Questions

What happens if AGENTS.md references a file that doesn't exist?

The resolve_imports function expects valid @path references within AGENTS.md. If a referenced file is missing, the script fails during the import resolution phase. Always verify that imported files exist at the specified paths before running the regeneration script.

Can I modify the generated instruction files directly?

No. The script prepends an "AUTO-GENERATED" comment header (lines 55-56) to warn against direct edits. Changes should be made to AGENTS.md or the imported files, then the script rerun to regenerate all platform-specific instruction files consistently.

How do I add support for a new AI agent?

Extend scripts/sync-agent-rules.sh by adding another write_file call (following the pattern in lines 69-86) pointing to your desired configuration file path. The script will automatically include the resolved content and auto-generated header in the new location.

Does the script handle nested directory structures?

Yes. The write_file function uses mkdir -p to create parent directories recursively before writing files. This ensures paths like .continue/rules/ are created automatically even if they did not exist previously.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}