Best Practices for Using Aider with the AI Website Cloner Template

Configure Aider with the repository's .aider.conf.yml file to restrict reads to AGENTS.md and docs/research/INSPECTION_GUIDE.md, and always execute scripts/sync-agent-rules.sh after updating canonical instructions to maintain consistency across all supported AI agents.

The JCodesMore/ai-website-cloner-template repository officially supports Aider as an AI coding agent, providing a minimal configuration file that whitelists specific source-of-truth documents. Following these best practices for using Aider with this template ensures reproducible website clones and prevents configuration drift across different agent implementations.

Configure the Aider Whitelist

The repository includes a .aider.conf.yml file at the root that explicitly defines which files Aider may read and edit. According to lines 1-3 of this configuration, the read: list is intentionally limited to two canonical documents:

read:
  - AGENTS.md
  - docs/research/INSPECTION_GUIDE.md

This restriction ensures that Aider stays synchronized with the central agent instructions and the inspection workflow guide. By limiting the read scope to these two files, you prevent accidental modifications to other repository files and ensure that generated skill files remain consistent across all supported agents.

Synchronize Configuration After Canonical Updates

When you modify AGENTS.md or add new instruction files, you must regenerate the platform-specific configurations. The repository provides scripts/sync-agent-rules.sh (referenced at line 21) to propagate these changes:

bash scripts/sync-agent-rules.sh

This script updates .aider.conf.yml and other agent-specific files to reflect the latest canonical instructions. Running this synchronization step is critical because it maintains the single source of truth for all agents, ensuring that Aider receives the same instructions as other AI coding agents in the ecosystem.

Launch Aider with the Repository Configuration

Always invoke Aider using the provided configuration file to enforce the read whitelist. Use this command from the repository root:

aider --config .aider.conf.yml

The --config flag guarantees that Aider respects the repository's file restrictions and does not attempt to read or modify files outside the defined read: list. Omitting this flag may result in Aider accessing files that should remain immutable according to the template's design.

Isolate Changes in New Repositories

Do not edit AGENTS.md or docs/research/INSPECTION_GUIDE.md directly within a fork that you intend to use for cloning a website. Instead, create a new repository from the template following the Quick-Start instructions (README lines 17-28).

Working in a fresh repository prevents accidental commits to the canonical files that could break the skill for other agents. This separation ensures that the original instruction files remain pristine while your specific website clone configurations reside in your isolated instance.

Verify Generated Skill Definitions

After running the sync script, inspect the generated skill definition at .github/skills/clone-website/SKILL.md to confirm that your changes to AGENTS.md have propagated correctly. This verification step ensures that the /clone-website skill Aider uses reflects the latest instructions before you begin the cloning process.

While Aider only reads .aider.conf.yml during execution, the actual skill logic resides in the .github/skills/clone-website/ directory. Confirming that this skill file aligns with your updated canonical documents prevents mismatches between the configuration whitelist and the available agent capabilities.

Summary

• Restrict Aider's scope via .aider.conf.yml to only read AGENTS.md and docs/research/INSPECTION_GUIDE.md
• Run the sync script (bash scripts/sync-agent-rules.sh) after any changes to canonical instruction files
• Launch with --config to ensure Aider respects the repository's file whitelist
• Create new repositories from the template rather than editing canonical files in forks
• Verify skill definitions at .github/skills/clone-website/SKILL.md after synchronization

Frequently Asked Questions

What files should Aider read in this template?

Aider should read only AGENTS.md and docs/research/INSPECTION_GUIDE.md as specified in .aider.conf.yml lines 1-3. These two files serve as the central source of truth for agent instructions and the inspection workflow guide, respectively. Limiting reads to these documents prevents drift and maintains consistency across all AI agents supported by the template.

How do I update Aider's configuration after modifying AGENTS.md?

Execute bash scripts/sync-agent-rules.sh from the repository root. This script, referenced at line 21, regenerates .aider.conf.yml and other agent configurations to reflect the latest changes in AGENTS.md. Running this script ensures that Aider receives the updated instructions without requiring manual configuration edits.

Can I edit the canonical instruction files directly in a fork?

No, you should not edit AGENTS.md or docs/research/INSPECTION_GUIDE.md directly in a fork intended for cloning websites. According to the README Quick-Start (lines 17-28), create a new repository from the template instead. This practice prevents accidental commits that could break the skill for other agents and keeps your website-specific changes isolated from the canonical instruction files.

Where is the actual skill definition stored for Aider?

The concrete skill definition resides in .github/skills/clone-website/SKILL.md, while Aider's runtime configuration reads from .aider.conf.yml. The skill file contains the detailed /clone-website implementation logic, whereas the configuration file simply dictates which canonical documents Aider may access. Verify the skill file after running the sync script to ensure it reflects the latest AGENTS.md changes.

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