research

Reusable baseline instructions for AI coding agents.

This repository is intended to be a small personal source of truth for AGENTS.md files. Copy the baseline into each project, then customize the project-specific sections.

Files

• AGENTS.md: Default baseline for most repositories.
• templates/project-AGENTS.md: Same baseline with explicit placeholders for project-specific notes.
• scripts/install-agents.ps1: Copies a template into a target project.

Usage

Copy the default baseline manually:

Copy-Item .\AGENTS.md C:\path\to\project\AGENTS.md

Or use the installer:

.\scripts\install-agents.ps1 -ProjectPath C:\path\to\project

Use the customizable template instead:

.\scripts\install-agents.ps1 -ProjectPath C:\path\to\project -Template .\templates\project-AGENTS.md

Customization Checklist

For each project, update:

• project overview
• repository structure
• common commands
• framework-specific rules
• security or deployment notes

Keep the shared baseline conservative. Put unusual rules in the target project's own AGENTS.md.

HTML Archive Framework

This repository includes the encrypted static HTML archive framework adapted from Lukael/research.

Typical report flow:

$env:REPORT_PASSWORD="<local secret>"
node scripts/build-markdown-report.js --slug example-report --input path\to\report.md --title "Example Report"

The command creates projects/<slug>/index.html and, when REPORT_PASSWORD is set, projects/<slug>/report.enc. The transient plaintext HTML is written under build/ and should not be committed.

Markdown reports render LaTeX math through the local KaTeX bundle under scripts/vendor/katex/. Use $...$ or \(...\) for inline math and $$...$$ or \[...\] for display math. Fenced code blocks remain literal and are not math-rendered.